{ "schema_version": "1.0.0", "license": "CC-BY-4.0", "attribution": "CanIEmbody (https://caniembody.pages.dev)", "dataset_version": "1.0.0-seed", "generated": "2026-10-25", "data": { "robots": [ "agilex-piper", "aloha", "arx5", "bi-so", "earthrover-mini-plus", "franka-fr3", "franka-panda", "hope-jr", "i2rt-yam", "koch", "lekiwi", "mobile-aloha", "omx", "openarm", "reachy2", "rebot-b601", "so100", "so101", "trossen-widowx", "unitree-g1", "ur5e", "widowx-250", "xarm", "xlerobot" ], "policies": [ "act", "diffusion", "eo1", "evo1", "fastwam", "groot-n1-7", "hil-serl", "lingbot-va", "molmoact2", "multitask-dit", "openvla-oft", "openvla", "pi0-fast", "pi0", "pi05", "smolvla", "tdmpc", "vla-jepa", "vqbet", "wall-oss", "xvla" ], "cells": [ { "robot": "agilex-piper", "policy": "act", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "LeRobot has no built-in Piper robot type; requires a third-party plugin (e.g. lerobot_robot_piper, robot.type=piper / piper_follower)." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-coport-act" ] }, "inference_real": { "state": "works-with-caveats", "evidence": [ "ev-issue-2942", "ev-plugin-agro-readme" ] } }, "evidence": [ { "id": "ev-hub-coport-act", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/coport-uni/piper_act_blue_200k", "date": "2026-02-21", "verified_on": "2026-10-24", "summary": "config.json type=act; trained on coport-uni/PiPER_pick_blue_colored_marker_to_box whose meta/info.json robot_type=piper_follower." }, { "id": "ev-issue-2942", "tier": "community-reported", "type": "github-issue", "url": "https://github.com/huggingface/lerobot/issues/2942", "date": "2026-02-10", "verified_on": "2026-10-24", "versions": { "lerobot": "0.4.3" }, "summary": "User controlling a Piper arm reports an ACT policy trained on their own dataset runs smoothly and completes the target tasks (contrast case for an X-VLA stutter bug).", "quote": "The same self-collected dataset was used to train an ACT model, which runs smoothly and successfully completes the target tasks." }, { "id": "ev-plugin-agro-readme", "tier": "community-reported", "type": "source-code", "url": "https://github.com/AgRoboticsResearch/lerobot_robot_piper/blob/abf721d9e88ac822b18cb3413a75bb8cc0e3b34f/README.md", "verified_on": "2026-10-24", "summary": "Third-party Piper plugin README documents ACT deployment on the Piper via LeRobot async inference (--robot.type=piper --policy_type=act).", "quote": "ACT Policy Deployment (Async Inference)" } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/agilex-piper/act" }, { "robot": "agilex-piper", "policy": "diffusion", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "The only physical run uses PCD-LeRobot, a research codebase with its own vendored LeRobot copy (pyproject version 0.3.4) and its own piper robot type; LeRobot has no built-in Piper robot type.", "The unmodified Diffusion Policy baseline succeeded 15% of the time on one bowl-on-plate task (50 demonstrations, one ROS RGB camera, 10 Hz inference); the trial count is not stated." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-aachen-diffusion", "ev-pcd-lerobot-piper-diffusion" ] }, "inference_real": { "state": "works-with-caveats", "notes": "15% baseline success on one task; trial count not stated.", "evidence": [ "ev-pcd-lerobot-piper-diffusion" ] }, "teleop_record": { "state": "works", "notes": "Leader-follower recording with --robot.type=piper_follower and --teleop.type=piper_leader in the PCD-LeRobot codebase.", "evidence": [ "ev-pcd-lerobot-piper-diffusion" ] } }, "evidence": [ { "id": "ev-hub-aachen-diffusion", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/RoboticsCollectiveAachen/diffusion_piper_hex_bolt_insertion", "date": "2026-10-20", "verified_on": "2026-10-24", "summary": "config.json type=diffusion; trained on Samisaliveagain/piper-hexagonal-bolt-insertion-100-demos-v1 whose meta/info.json robot_type=piper_follower. Card states it has not yet been evaluated on the robot." }, { "id": "ev-pcd-lerobot-piper-diffusion", "tier": "community-reported", "type": "source-code", "url": "https://github.com/pcd-robot/PCD-LeRobot/blob/519b4a814e85bf9b786677d90b0ff07218729bb2/README.md", "date": "2025-11-25", "verified_on": "2026-10-25", "versions": { "lerobot": "0.3.4 (vendored copy in PCD-LeRobot)" }, "summary": "README of the Policy Contrastive Decoding (PCD) authors' LeRobot-based code gives AgileX PiPER recording, Diffusion Policy training and 10 Hz on-robot inference commands; its real-robot results table reports 15% success for the plain Diffusion Policy (45% with PCD), without a trial count.", "quote": "We designed a \"place bowl on plate\" task using an AGILEX PIPER robotic arm, collected 50 demonstration trajectories and trained a Diffusion Policy from scratch." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/agilex-piper/diffusion" }, { "robot": "agilex-piper", "policy": "lingbot-va", "state": "partial", "partial_scope": "Training evidenced by a public checkpoint (upstream LingBot-VA runtime plus a local patch); only offline validation is documented, no on-robot inference.", "framework": "lingbot-va", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-wkrfugs-lingbot" ] } }, "evidence": [ { "id": "ev-hub-wkrfugs-lingbot", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/wkrfugs/lingbot-va-piper-4task-step30000", "date": "2026-10-20", "verified_on": "2026-10-24", "summary": "Model card: robbyant/lingbot-va-base fine-tuned on four single-arm Piper tasks for 30k steps using the official Robbyant/lingbot-va runtime with a published patch; card describes offline validation and states the adapter is not a robot client or controller." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/agilex-piper/lingbot-va" }, { "robot": "agilex-piper", "policy": "pi0", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-coport-pi0" ] } }, "evidence": [ { "id": "ev-hub-coport-pi0", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/coport-uni/piper_pi0_blue_20k", "date": "2026-02-25", "verified_on": "2026-10-24", "summary": "config.json type=pi0; trained on coport-uni/PiPER_pick_blue_colored_marker_to_box whose meta/info.json robot_type=piper_follower." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/agilex-piper/pi0" }, { "robot": "agilex-piper", "policy": "pi05", "state": "works-with-caveats", "framework": "openpi", "caveats": [ "The vendor-documented route is AgileX's own openpi fork (openpi-agilex, JAX, ROS 2) on the bimanual Cobot Magic (two PiPER arms, three cameras), not a single PiPER arm and not LeRobot.", "AgileX documents the fine-tuning and inference workflow but publishes no PiPER success rates or released PiPER checkpoint." ], "capabilities": { "train": { "state": "works", "notes": "Vendor: π0.5 fine-tuning configs pi05_aloha / pi05_aloha_lora for cobot_magic data collected with agilexrobotics/data_tools. Community: LeRobot pi05 checkpoint trained on single-arm Piper data.", "evidence": [ "ev-openpi-agilex-readme", "ev-hub-qualia-pi05" ] }, "inference_real": { "state": "works", "notes": "Vendor: serve the fine-tuned checkpoint with openpi's serve_policy.py and run scripts/aloha_inference-joint-ros2.py against dual PiPER arms.", "evidence": [ "ev-openpi-agilex-readme", "ev-openpi-agilex-configs" ] } }, "evidence": [ { "id": "ev-openpi-agilex-readme", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/agilexrobotics/openpi-agilex/blob/432577c37c4fc650fcbea03f03e45f936e9e4a08/README.md", "date": "2026-02-02", "verified_on": "2026-10-25", "summary": "AgileX's openpi distribution README: fine-tune π0.5 (pi05_base) on your data, serve it with serve_policy.py, then run inference on dual PiPER arms via the provided ROS 2 scripts.", "quote": "open your dual piper arms ros-sdk and multi cameras, then" }, { "id": "ev-openpi-agilex-configs", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/agilexrobotics/openpi-agilex/blob/432577c37c4fc650fcbea03f03e45f936e9e4a08/README.md", "date": "2026-02-02", "verified_on": "2026-10-25", "summary": "Same README's config table maps the π0.5 configs pi05_aloha and pi05_aloha_lora to the inference script for the Cobot Magic (three cameras, two arms).", "quote": "| pi05_aloha | scripts/aloha_inference-joint-ros2.py | cobot_magic, 3 cameras and 2 arms |" }, { "id": "ev-hub-qualia-pi05", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/qualia-robotics/pi05-piper-popcorn-6956e853", "date": "2026-03-26", "verified_on": "2026-10-24", "summary": "config.json type=pi05; trained on callum-sh/piper-popcorn whose meta/info.json robot_type=piper." } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/agilex-piper/pi05" }, { "robot": "agilex-piper", "policy": "smolvla", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "With async inference the arm swayed back and forth as the action queue refilled; the reporter fixed it by tuning the queue size (maintainer suggested RTC).", "LeRobot has no built-in Piper robot type; requires a third-party plugin." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-wego-smolvla" ] }, "inference_real": { "state": "works-with-caveats", "evidence": [ "ev-issue-2938" ] } }, "evidence": [ { "id": "ev-hub-wego-smolvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/wego-hansu/piper_smolvla", "date": "2025-10-11", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; trained on wego-hansu/piper_pick_cars_033 whose meta/info.json robot_type=piper_follower." }, { "id": "ev-issue-2938", "tier": "community-reported", "type": "github-issue", "url": "https://github.com/huggingface/lerobot/issues/2938", "date": "2026-02-09", "verified_on": "2026-10-24", "versions": { "lerobot": "0.4.3" }, "summary": "User (Robot Arm: PiPER) ran a fine-tuned SmolVLA with async inference; motion swayed as the queue refilled, improved after adjusting queue size.", "quote": "I fine-tuned the SmolVLA model to perform asynchronous inference." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/agilex-piper/smolvla" }, { "robot": "agilex-piper", "policy": "vqbet", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-kimeuna-vqbet" ] } }, "evidence": [ { "id": "ev-hub-kimeuna-vqbet", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/KimEuna/piper_wetwipe_test_vqbet_Final", "date": "2026-03-20", "verified_on": "2026-10-24", "summary": "config.json type=vqbet; trained on jj-jjj/piper_wetwipe_test whose meta/info.json robot_type=piper_follower." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/agilex-piper/vqbet" }, { "robot": "agilex-piper", "policy": "wall-oss", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-jolch-wallx" ] } }, "evidence": [ { "id": "ev-hub-jolch-wallx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/jolch/piper_wall_x_20260207", "date": "2026-02-07", "verified_on": "2026-10-24", "summary": "config.json type=wall_x (LeRobot WALL-OSS); trained on jolch/piper_pnp_20260111_1538 whose meta/info.json robot_type=piper_follower." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/agilex-piper/wall-oss" }, { "robot": "agilex-piper", "policy": "xvla", "state": "works-with-caveats", "framework": "x-vla", "caveats": [ "Upstream evidence is the bimanual AgileX Aloha platform (two PiPER arms driven through AgileX's Piper ROS 1 Noetic driver) using X-VLA's own server-client deployment, not a single PiPER arm and not LeRobot.", "Through LeRobot, open issue #2942 (two users): X-VLA on a Piper moves intermittently with pauses because action outputs change step-wise; ACT on the same data runs smoothly." ], "capabilities": { "train": { "state": "works", "notes": "Upstream: X-VLA-0.9B fine-tuned on Soft-Fold, 1,200 bimanual AgileX cloth-folding trajectories (released as 2toINF/X-VLA-SoftFold). Community: LeRobot X-VLA checkpoint on single-arm piper_follower data.", "evidence": [ "ev-xvla-paper-softfold", "ev-hub-vis22-xvla" ] }, "inference_real": { "state": "works-with-caveats", "notes": "Upstream: real-world cloth folding on the bimanual AgileX Aloha (about one cloth folded in under two minutes). Through LeRobot on a single Piper: runs but stutters (issue #2942, unresolved).", "evidence": [ "ev-xvla-paper-softfold-result", "ev-xvla-softfold-agilex-readme", "ev-issue-2942" ] } }, "evidence": [ { "id": "ev-hub-vis22-xvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/vis22/xvla-piper-robot", "date": "2026-10-11", "verified_on": "2026-10-24", "summary": "config.json type=xvla (base_model lerobot/xvla-base); trained on vis22/green_yellow_plates whose meta/info.json robot_type=piper_follower." }, { "id": "ev-issue-2942", "tier": "community-reported", "type": "github-issue", "url": "https://github.com/huggingface/lerobot/issues/2942", "date": "2026-02-10", "verified_on": "2026-10-24", "versions": { "lerobot": "0.4.3" }, "summary": "Bug report: X-VLA controlling a Piper arm produces stuttering motion; a second user reports the same; open, no fix.", "quote": "When using the XVLA model to control the Piper robotic arm, the arm's movement exhibits noticeable stuttering (i.e., moving intermittently with pauses)." }, { "id": "ev-xvla-paper-softfold", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2510.10274", "date": "2025-10-11", "verified_on": "2026-10-25", "summary": "X-VLA paper, Sec. 5: the Soft-Fold cloth-folding dataset is collected on the bimanual AgileX platform and X-VLA-0.9B is fine-tuned on it (Appendix F).", "quote": "we build a high-quality cloth-folding dataset on the bi-manual Agilex platform, termed Soft-Fold, which consists of 1,200 trajectories collected through a carefully designed pipeline." }, { "id": "ev-xvla-paper-softfold-result", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2510.10274", "date": "2025-10-11", "verified_on": "2026-10-25", "summary": "X-VLA paper abstract: with the 1,200 Soft-Fold demonstrations X-VLA performs real-world dexterous cloth folding; Appendix J lists AgileX among its three real-world embodiments.", "quote": "with only 1,200 demonstrations, it excels at a dexterous cloth-folding manipulation task in the real world, achieving an average throughput of folding 1 cloth under two minutes." }, { "id": "ev-xvla-softfold-agilex-readme", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/2toinf/X-VLA/blob/6c4ecfb8be458ae7c0c96094e4b0add4174f4fe2/evaluation/SoftFold-Agilex/readme.md", "date": "2025-11-06", "verified_on": "2026-10-25", "summary": "X-VLA repo evaluation/SoftFold-Agilex: serve 2toINF/X-VLA-SoftFold with deploy.py and run the client deploy/client_eef6d_xvla.py against the AgileX Aloha, whose arms are launched with the bundled Piper ROS driver (roslaunch piper start_ms_piper.launch, piper_description URDF).", "quote": "We evaluate **X-VLA** on the **Agilex Aloha platform** to perform long-horizon cloth-folding tasks." } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/agilex-piper/xvla" }, { "robot": "aloha", "policy": "act", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "LeRobot's built-in `aloha` robot type was removed in the hardware API redesign (PR #777, June 2025) and is not in v0.6.1; the documented LeRobot workflow needs a pre-redesign commit or a third-party driver.", "ACT was introduced on the original ALOHA (ViperX 300 6-DoF followers) using the authors' own codebase (tonyzhaozh/act), not LeRobot." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-lerobot-aloha-doc", "ev-trossen-lerobot-guide", "ev-act-paper", "ev-hub-trossen-act-tshirt" ] }, "inference_real": { "state": "works", "evidence": [ "ev-act-paper", "ev-lerobot-aloha-doc", "ev-trossen-lerobot-guide" ] }, "teleop_record": { "state": "works", "evidence": [ "ev-lerobot-aloha-doc", "ev-trossen-lerobot-guide" ] }, "inference_sim": { "state": "works", "notes": "Simulated ALOHA (MuJoCo / gym-aloha) only.", "evidence": [ "ev-act-paper", "ev-lerobot-act-aloha-sim" ] } }, "evidence": [ { "id": "ev-act-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2304.13705", "date": "2023-04-23", "verified_on": "2026-10-24", "summary": "The ACT paper introduces both ALOHA and ACT, and evaluates ACT on 6 real-world ALOHA tasks plus 2 simulated (MuJoCo) ALOHA tasks.", "quote": "We present experiments to evaluate ACT’s performance on fine manipulation tasks. For ease of reproducibility, we build two simulated fine manipulation tasks in MuJoCo [63], in addition to 6 real-world tasks with ALOHA." }, { "id": "ev-lerobot-aloha-doc", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/bed90e3a41c43758c619dba66158ddd5798d361a/examples/9_use_aloha.md", "verified_on": "2026-10-24", "summary": "Official LeRobot ALOHA tutorial (last version before removal, commit bed90e3, May 2025): records with --robot.type=aloha, trains with --policy.type=act, and evaluates the ACT checkpoint on the robot via control_robot.py record --control.policy.path.", "quote": "This tutorial explains how to use [Aloha and Aloha 2 stationary](https://www.trossenrobotics.com/aloha-stationary) with LeRobot." }, { "id": "ev-trossen-lerobot-guide", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://docs.trossenrobotics.com/aloha_docs/2.0/training/lerobot_guide.html", "verified_on": "2026-10-24", "summary": "Trossen Robotics' 'LeRobot X Aloha User Guide' (ALOHA kits with ViperX 300 S followers) covers teleop, recording, ACT training (policy=act_aloha_real) and on-robot evaluation, pinned to LeRobot commit 96c7052777aca85d4e55dfba8f81586103ba8f61.", "quote": "The policy is specified with policy=act_aloha_real." }, { "id": "ev-hub-trossen-act-tshirt", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/TrossenRoboticsCommunity/act-fold-tshirt-aloha", "verified_on": "2026-10-24", "summary": "config.json type=act; trained on TrossenRoboticsCommunity/aloha_fold_tshirt whose meta/info.json robot_type=aloha (21 episodes, 18-dim joint state incl. shadow joints, 3 cameras). Card tags platform:aloha, robot:stationary." }, { "id": "ev-lerobot-act-aloha-sim", "tier": "upstream-documented", "type": "model-card", "url": "https://huggingface.co/lerobot/act_aloha_sim_transfer_cube_human", "verified_on": "2026-10-24", "summary": "Official LeRobot checkpoint: config.json type=act; train_config.json dataset lerobot/aloha_sim_transfer_cube_human, env type aloha, task AlohaTransferCube-v0 (simulation, gym-aloha).", "quote": "trained for the `AlohaTransferCube` environment from [gym-aloha](https://github.com/huggingface/gym-aloha)." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/aloha/act" }, { "robot": "aloha", "policy": "diffusion", "state": "works-with-caveats", "framework": "droid-policy-learning", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "Run by the OpenVLA-OFT authors as a from-scratch baseline using the DROID Diffusion Policy implementation (language-conditioned, modified for bimanual control), not LeRobot's diffusion policy.", "Struggled on the 'put X into pot' task, which had the largest training set." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-oft-paper" ] }, "inference_real": { "state": "works", "evidence": [ "ev-oft-paper" ] } }, "evidence": [ { "id": "ev-oft-paper", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2502.19645", "date": "2025-02-27", "verified_on": "2026-10-24", "summary": "OpenVLA-OFT paper trains Diffusion Policy from scratch on four real tasks on the original ALOHA (two ViperX 300 S arms) and reports it matching or exceeding RDT-1B on clothes folding and scooping.", "quote": "Diffusion Policy demonstrates stronger capabilities, matching or exceeding RDT-1B’s reliability on the clothes folding and scooping tasks. However, it struggles with the “put X into pot” task which has a larger training dataset" } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/aloha/diffusion" }, { "robot": "aloha", "policy": "evo1", "state": "partial", "partial_scope": "The upstream Evo-1 repo ships a community-contributed real-ALOHA inference client (ROS1 Noetic, Interbotix); no ALOHA training recipe, checkpoint or on-robot results are documented.", "framework": "evo-1", "capabilities": { "inference_real": { "state": "partial", "notes": "Client/server inference scripts only.", "evidence": [ "ev-evo1-readme", "ev-evo1-aloha-readme" ] } }, "evidence": [ { "id": "ev-evo1-readme", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/MINT-SJTU/Evo-1/blob/5fd14b015013c4fd0aacf5f8f48f868ca9b870a2/README.md", "date": "2025-12-15", "verified_on": "2026-10-24", "summary": "Evo-1 README news entry (2025-12-15) announcing ALOHA dual-arm inference code contributed by a community user.", "quote": "Added Evo-1 inference code in Aloha dual arm (Implemented by community user @meijie-jesse)" }, { "id": "ev-evo1-aloha-readme", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/MINT-SJTU/Evo-1/blob/5fd14b015013c4fd0aacf5f8f48f868ca9b870a2/Evo_1/scripts/README_Aloha.md", "verified_on": "2026-10-24", "summary": "Run instructions for Evo1_server.py + Evo1_client_aloha.py on a real ALOHA built per tonyzhaozh/aloha (robot_utils.py uses interbotix_xs_msgs).", "quote": "This example demonstrates how to run EVO1 with a real robot using an ALOHA setup." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/aloha/evo1" }, { "robot": "aloha", "policy": "openvla-oft", "state": "works", "framework": "openvla-oft", "caveats": [ "Real-robot results use the OFT+ variant (OFT plus FiLM language conditioning) on the original ALOHA (two ViperX 300 S arms, 3 cameras), run at 25 Hz instead of ALOHA's native 50 Hz." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-oft-paper", "ev-oft-aloha-md" ] }, "inference_real": { "state": "works", "evidence": [ "ev-oft-paper", "ev-oft-aloha-md" ] } }, "evidence": [ { "id": "ev-oft-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2502.19645", "date": "2025-02-27", "verified_on": "2026-10-24", "summary": "OpenVLA-OFT paper fine-tunes OpenVLA with OFT+ on four real ALOHA tasks (fold shorts, fold shirt, scoop X into bowl, put X into pot); ALOHA here is the original ALOHA of Zhao et al. 2023 (ref. [56], arXiv 2304.13705), not ALOHA 2.", "quote": "The ALOHA platform comprises two ViperX 300 S arms, three camera viewpoints (one top-down, two wrist-mounted), and robot state inputs (14-dimensional joint angles). It operates at 25 Hz (reduced from the original 50 Hz to enable faster training while still maintaining smooth robotic control)" }, { "id": "ev-oft-aloha-md", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/moojink/openvla-oft/blob/e4287e94541f459edc4feabc4e181f537cd569a8/ALOHA.md", "verified_on": "2026-10-24", "summary": "Upstream repo documents ALOHA data preprocessing, fine-tuning (vla-scripts/finetune.py) and server-client real-robot evaluation (run_aloha_eval.py + deploy.py).", "quote": "# OpenVLA-OFT+ in Real-World ALOHA Robot Tasks" } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/aloha/openvla-oft" }, { "robot": "aloha", "policy": "openvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "openvla", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-anjali-openvla-lora", "ev-hub-anjali-openvla-full" ] } }, "evidence": [ { "id": "ev-hub-anjali-openvla-lora", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/anjali209/openvla-7b-aloha-coffee-lora", "date": "2026-08-01", "verified_on": "2026-10-25", "summary": "Model card: LoRA (r=32) on openvla/openvla-7b fine-tuned on 48 episodes of lerobot/aloha_static_coffee (meta/info.json robot_type=aloha, real static ALOHA, 50 episodes), using only cam_left_wrist and 14-D joint targets discretised into 256 bins. Card states it is not validated on a robot or in simulation and fails both held-out episodes offline. Not a LeRobot-format checkpoint (no config.json policy type)." }, { "id": "ev-hub-anjali-openvla-full", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/anjali209/openvla-7b-aloha-coffee-full", "date": "2026-08-01", "verified_on": "2026-10-25", "summary": "Model card: full fine-tune of openvla/openvla-7b on the same 48 lerobot/aloha_static_coffee episodes (single wrist camera); offline teacher-forced evaluation only, same stated limitations." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/aloha/openvla" }, { "robot": "aloha", "policy": "pi0-fast", "state": "works-with-caveats", "framework": "openpi", "caveats": [ "Evidence is Physical Intelligence's own zero-shot evaluation of its generalist π0-FAST on an in-house bimanual Trossen ViperX setup that mirrors ALOHA; openpi ships no π0-FAST ALOHA config or checkpoint." ], "capabilities": { "inference_real": { "state": "works", "evidence": [ "ev-fast-paper" ] } }, "evidence": [ { "id": "ev-fast-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2501.09747", "date": "2025-01-16", "verified_on": "2026-10-24", "summary": "FAST paper: the generalist π0-FAST is evaluated zero-shot on the 'toast out of toaster' task, run on a bimanual Trossen ViperX robot mirroring the ALOHA setup (Section VI-F, Fig. 11).", "quote": "This task requires a bi-manual Trossen ViperX robot, mirroring the ALOHA [70] setup, to take two pieces of toast out of a toaster and place them onto a plate. This task was evaluated out-of-the-box on models pretrained with the full mixture detailed in Black et al. [7]." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/aloha/pi0-fast" }, { "robot": "aloha", "policy": "pi0", "state": "works", "framework": "openpi", "caveats": [ "openpi's ALOHA checkpoints were fine-tuned on Physical Intelligence's ALOHA data; upstream warns zero-shot use is experimental and recommends fine-tuning π0 on data from the target robot." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-openpi-aloha-real" ] }, "inference_real": { "state": "works", "evidence": [ "ev-openpi-readme", "ev-openpi-aloha-real" ] } }, "evidence": [ { "id": "ev-openpi-readme", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/README.md", "verified_on": "2026-10-24", "summary": "openpi lists fine-tuned π0 'expert' checkpoints for the ALOHA platform (pi0_aloha_towel, pi0_aloha_tupperware, pi0_aloha_pen_uncap).", "quote": "$\\pi_0$ model fine-tuned on internal [ALOHA](https://tonyzhaozh.github.io/aloha/) data: can fold diverse towels 0-shot on ALOHA robot platforms" }, { "id": "ev-openpi-aloha-real", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/examples/aloha_real/README.md", "verified_on": "2026-10-24", "summary": "openpi's real-ALOHA example: runs π0 checkpoints on an ALOHA setup (tonyzhaozh/aloha hardware, ROS) via a policy server, and documents fine-tuning on your own ALOHA dataset (pi0_aloha_pen_uncap config).", "quote": "This example demonstrates how to run with a real robot using an [ALOHA setup](https://github.com/tonyzhaozh/aloha)." }, { "id": "ev-pi0-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2410.24164", "date": "2024-10-31", "verified_on": "2026-10-24", "summary": "π0 paper: one of the training/evaluation platforms is a bimanual Trossen ViperX setup based on ALOHA.", "quote": "Bimanual Trossen. This setup has two 6-DoF Trossen ViperX arms in a configuration based on the ALOHA setup [4, 57], with two wrist cameras and a base camera, and a 14-dimensional configuration and action space." }, { "id": "ev-hub-trossen-pi0-tshirt", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/TrossenRoboticsCommunity/pi0-fold-tshirt-aloha", "verified_on": "2026-10-24", "summary": "LeRobot-format checkpoint: config.json type=pi0; trained on TrossenRoboticsCommunity/aloha_fold_tshirt whose meta/info.json robot_type=aloha (21 episodes). Card tags platform:aloha, robot:stationary, framework:lerobot-latest." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/aloha/pi0" }, { "robot": "aloha", "policy": "pi05", "state": "partial", "partial_scope": "openpi ships π0.5 ALOHA configs (inference config pi05_aloha and fine-tuning config pi05_aloha_pen_uncap on public ALOHA data), but no π0.5 ALOHA checkpoint and no published on-robot ALOHA result.", "framework": "openpi", "capabilities": { "train": { "state": "works", "notes": "Upstream fine-tuning config on physical-intelligence/aloha_pen_uncap_diverse from pi05_base.", "evidence": [ "ev-openpi-config" ] } }, "evidence": [ { "id": "ev-openpi-config", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/src/openpi/training/config.py", "verified_on": "2026-10-24", "summary": "openpi training configs include pi05_aloha (under '# Inference Aloha configs.') and pi05_aloha_pen_uncap, a π0.5 fine-tuning config on the public ALOHA pen-uncap dataset with trossen normalization assets.", "quote": "name=\"pi05_aloha_pen_uncap\"," } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/aloha/pi05" }, { "robot": "aloha", "policy": "smolvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-qualia-smolvla-coffee", "ev-hub-qualia-smolvla-cups" ] } }, "evidence": [ { "id": "ev-hub-qualia-smolvla-coffee", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/qualia-robotics/smolvla-aloha-static-coffee-87e074e8", "verified_on": "2026-10-24", "summary": "config.json type=smolvla (base lerobot/smolvla_base); train_config.json dataset lerobot/aloha_static_coffee, whose meta/info.json robot_type=aloha (50 real static-ALOHA episodes, 4 cameras, 14-dim joint state/action). Template card, no rollout results." }, { "id": "ev-hub-qualia-smolvla-cups", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/qualia-robotics/smolvla-aloha-static-cups-open-bd3cb6ef", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config.json dataset lerobot/aloha_static_cups_open, whose meta/info.json robot_type=aloha (50 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/aloha/smolvla" }, { "robot": "arx5", "policy": "act", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-villekuosmanen-act" ] } }, "evidence": [ { "id": "ev-hub-villekuosmanen-act", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/villekuosmanen/build_block_tower_1.0.0", "date": "2025-12-10", "verified_on": "2026-10-25", "summary": "config.json type=act; train_config.json dataset villekuosmanen/build_block_tower (60000 steps), whose meta/info.json robot_type=arx5 (100 episodes, 7-D joint state, front + wrist cameras). The string is arx5, not the plugin's arx5_follower; the dataset author maintains the lerobot-arx5 plugin. Auto-generated card, no results." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/arx5/act" }, { "robot": "arx5", "policy": "diffusion", "state": "works-with-caveats", "framework": "umi-on-legs", "caveats": [ "Third-party research paper (UMI on Legs, from the Diffusion Policy authors' lab), not documented in Diffusion Policy's own docs or in LeRobot; the ARX5 LeRobot plugin documents no Diffusion Policy run.", "The ARX5 was mounted on a Unitree Go2 quadruped and driven through a learned whole-body controller; policies were trained on handheld UMI gripper demonstrations (or the public UMI cup checkpoint), not on ARX5 robot data." ], "capabilities": { "train": { "state": "works", "notes": "LeRobot diffusion policy trained on ARX5 robot data (public checkpoint), separate from the UMI-on-Legs evidence.", "evidence": [ "ev-hub-villekuosmanen-diffusion" ] }, "inference_real": { "state": "works", "notes": "Diffusion Policy inference on a desktop RTX 4090 over the network; tasks: dynamic tossing, kettlebell pushing and in-the-wild cup rearrangement (zero-shot with the public UMI checkpoint).", "evidence": [ "ev-umi-on-legs-system", "ev-umi-on-legs-policy" ] } }, "evidence": [ { "id": "ev-umi-on-legs-system", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2407.10353v1", "date": "2024-07-14", "verified_on": "2026-10-25", "summary": "UMI on Legs, Sec. 3.3: the robot is a Unitree Go2 with a 6-DOF ARX5 arm; diffusion policy inference runs on a separate desktop RTX 4090.", "quote": "consists of a 12-DOF Unitree Go2 quadruped robot and a 6-DOF ARX5 robot arm, both powered by the battery of Go2." }, { "id": "ev-umi-on-legs-policy", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2407.10353v1", "date": "2024-07-14", "verified_on": "2026-10-25", "summary": "UMI on Legs, Fig. 3: the manipulation policy is a diffusion policy (Chi et al. default U-Net configuration) trained on real-world UMI demonstrations, whose end-effector trajectory drives the whole-body controller.", "quote": "infers a camera-frame end-effector trajectory using a diffusion policy (a), trained using real-world UMI demonstrations." }, { "id": "ev-hub-villekuosmanen-diffusion", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/villekuosmanen/build_block_tower_diffusion", "date": "2026-03-23", "verified_on": "2026-10-25", "summary": "config.json type=diffusion; train_config.json dataset villekuosmanen/build_block_tower (60000 steps), whose meta/info.json robot_type=arx5 (100 episodes, 7-D joint state, front + wrist cameras). The string is arx5, not the plugin's arx5_follower; the dataset author maintains the lerobot-arx5 plugin. Auto-generated card, no results." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/arx5/diffusion" }, { "robot": "arx5", "policy": "pi0", "state": "works-with-caveats", "framework": "openpi", "caveats": [ "Community-reported only (RoboChallenge benchmark by Dexmal/Hugging Face); not documented by Physical Intelligence or maintainer-tested.", "The paper does not name the codebase; openpi is recorded because RoboChallenge maintains an openpi fork and π0 is open-sourced there.", "Per-task fine-tuned π0 succeeded unevenly on ARX5 tasks, e.g. 80% on place shoes on rack but 0% on open the drawer and fold dishcloth." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-robochallenge-hw", "ev-robochallenge-methods" ] }, "inference_real": { "state": "works-with-caveats", "evidence": [ "ev-robochallenge-hw", "ev-robochallenge-methods" ] } }, "evidence": [ { "id": "ev-robochallenge-hw", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2510.17950", "date": "2025-10-20", "verified_on": "2026-10-24", "summary": "RoboChallenge (Table30) real-robot benchmark fleet includes a table-mounted ARX-5 arm; 10 of the 30 tasks run on ARX5.", "quote": "ARX-5 arm. A 6-DOF ARX-5 arm, mounted on a table." }, { "id": "ev-robochallenge-methods", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2510.17950", "date": "2025-10-20", "verified_on": "2026-10-24", "summary": "π0 is one of the baselines fine-tuned and evaluated on all 30 Table30 tasks, including the ARX5 tasks (results table: Pi0 column).", "quote": "As our initial survey, we tested four popular open source VLA algorithms:" } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/arx5/pi0" }, { "robot": "arx5", "policy": "pi05", "state": "works-with-caveats", "framework": "openpi", "caveats": [ "Community-reported only (RoboChallenge benchmark by Dexmal/Hugging Face, plus a community checkpoint); not documented by Physical Intelligence or maintainer-tested.", "Per-task fine-tuned π0.5 succeeded unevenly on ARX5 tasks, e.g. 90% on place shoes on rack and put cup on coaster but 0% on arrange paper cups." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-robochallenge-methods", "ev-hub-pravsels-pi05" ] }, "inference_real": { "state": "works-with-caveats", "evidence": [ "ev-robochallenge-hw", "ev-robochallenge-methods" ] } }, "evidence": [ { "id": "ev-robochallenge-hw", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2510.17950", "date": "2025-10-20", "verified_on": "2026-10-24", "summary": "RoboChallenge (Table30) real-robot benchmark fleet includes a table-mounted ARX-5 arm; 10 of the 30 tasks run on ARX5.", "quote": "ARX-5 arm. A 6-DOF ARX-5 arm, mounted on a table." }, { "id": "ev-robochallenge-methods", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2510.17950", "date": "2025-10-20", "verified_on": "2026-10-24", "summary": "π0.5 is one of the baselines fine-tuned and evaluated on all 30 Table30 tasks including the ARX5 tasks; it was the strongest model overall (43.7% average success).", "quote": "As our initial survey, we tested four popular open source VLA algorithms:" }, { "id": "ev-hub-pravsels-pi05", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/pravsels/pi05-arx5-multitask-v1", "date": "2026-03-24", "verified_on": "2026-10-24", "summary": "Model card: π0.5 (openpi, Orbax checkpoints) fine-tuned from pi05_base on 186 ARX5 single-arm and bimanual LeRobot v2.1 datasets; training-loss table only, no robot eval." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/arx5/pi05" }, { "robot": "bi-so", "policy": "act", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-pepijn-act-biso100", "ev-hub-nbirukov-act-biso101" ] } }, "evidence": [ { "id": "ev-hub-pepijn-act-biso100", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/pepijn223/bimanual-so100-handover-cube", "date": "2025-07-16", "verified_on": "2026-10-24", "summary": "config.json type=act; train_config.json dataset.repo_id=pepijn223/bimanual-so100-handover-cube, whose meta/info.json robot_type=bi_so100_follower (25 episodes)." }, { "id": "ev-hub-nbirukov-act-biso101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/nbirukov/act_bimanual-so101-stack-rings-fast", "date": "2025-10-06", "verified_on": "2026-10-24", "summary": "config.json type=act; train_config.json dataset.repo_id=nbirukov/bimanual-so101-stack-yellow-ring-fast-V2, whose meta/info.json robot_type=bi_so101_follower (300 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/bi-so/act" }, { "robot": "bi-so", "policy": "diffusion", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-thomas-diffusion-biso-push", "ev-hub-thomas-diffusion-biso-stack" ] } }, "evidence": [ { "id": "ev-hub-thomas-diffusion-biso-push", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/thomas0829/policy-bimanual-diffusion-push-and-grab", "date": "2025-11-17", "verified_on": "2026-10-24", "summary": "config.json type=diffusion; train_config.json dataset.repo_id=thomas0829/bimanual_so101_push_and_grab, whose meta/info.json robot_type=bi_so100_follower (30 episodes)." }, { "id": "ev-hub-thomas-diffusion-biso-stack", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/thomas0829/policy-bimanual-diffusion-stacking", "date": "2025-11-11", "verified_on": "2026-10-24", "summary": "config.json type=diffusion; train_config.json dataset.repo_id=thomas0829/bimanual-so101-stacking-blocks-v2, whose meta/info.json robot_type=bi_so100_follower (50 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/bi-so/diffusion" }, { "robot": "bi-so", "policy": "groot-n1-7", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-orderdraconis-groot-v2", "ev-hub-orderdraconis-groot-leo" ] } }, "evidence": [ { "id": "ev-hub-orderdraconis-groot-v2", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/OrderDraconis/groot_phase2align_v2_c30", "date": "2026-10-11", "verified_on": "2026-10-25", "summary": "config.json type=groot with base_model_path nvidia/GR00T-N1.7-3B; train_config.json dataset OrderDraconis/phase2_align_combined_full_V2 (40000 steps), whose meta/info.json robot_type=bi_so_follower (249 episodes). The card only gives a templated lerobot-rollout command for bi_so_follower, no results." }, { "id": "ev-hub-orderdraconis-groot-leo", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/OrderDraconis/groot_align_leo", "date": "2026-08-24", "verified_on": "2026-10-25", "summary": "config.json type=groot with base_model_path nvidia/GR00T-N1.7-3B; train_config.json dataset Janmeier820/align_fabric_dataset_combined (10000 steps), whose meta/info.json robot_type=bi_so_follower (173 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/bi-so/groot-n1-7" }, { "robot": "bi-so", "policy": "molmoact2", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-mrc4t-molmoact2" ] } }, "evidence": [ { "id": "ev-hub-mrc4t-molmoact2", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/MrC4t/molmoact2_bi_so_bin5", "date": "2026-09-19", "verified_on": "2026-10-25", "summary": "config.json type=molmoact2; train_config.json dataset MrC4t/bimanual_toy_bin (2000 steps), whose meta/info.json robot_type=bi_so_follower (46 episodes). Card has no rollout results." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/bi-so/molmoact2" }, { "robot": "bi-so", "policy": "multitask-dit", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-chihhan-dit-biso101", "ev-hub-notstrober-dit-biso" ] } }, "evidence": [ { "id": "ev-hub-chihhan-dit-biso101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/ChihHanShen/multitask-dit-so101", "date": "2026-09-08", "verified_on": "2026-10-24", "summary": "config.json type=multi_task_dit; train_config.json dataset.repo_id=ChihHanShen/bimanual-so101-pickvials-real, whose meta/info.json robot_type=bi_so101_follower (101 episodes)." }, { "id": "ev-hub-notstrober-dit-biso", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/notstrober/desk-stationery-to-drawer-bimanual_v2-multitask-dit-bs24-50k", "date": "2026-05-14", "verified_on": "2026-10-24", "summary": "config.json type=multi_task_dit; train_config.json dataset.repo_id=notstrober/desk-stationery-to-drawer-bimanual_v2, whose meta/info.json robot_type=bi_so_follower (356 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/bi-so/multitask-dit" }, { "robot": "bi-so", "policy": "pi0-fast", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-delvingdeep-pi0fast-biso101" ] } }, "evidence": [ { "id": "ev-hub-delvingdeep-pi0fast-biso101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/delvingdeep/pi0fast-so101-bimanual", "date": "2026-02-01", "verified_on": "2026-10-24", "summary": "config.json type=pi0_fast; train_config.json dataset.repo_id=Harshita979797/so101_bimanual_dataset, whose meta/info.json robot_type=bi_so101_follower (43 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/bi-so/pi0-fast" }, { "robot": "bi-so", "policy": "pi0", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-pepijn-pi0-biso100", "ev-hub-wuc1-pi0-biso" ] } }, "evidence": [ { "id": "ev-hub-pepijn-pi0-biso100", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/pepijn223/bimanual-so100-handover-cube_pi0", "date": "2025-10-07", "verified_on": "2026-10-24", "summary": "config.json type=pi0; train_config.json dataset.repo_id=pepijn223/bimanual-so100-handover-cube, whose meta/info.json robot_type=bi_so100_follower (25 episodes)." }, { "id": "ev-hub-wuc1-pi0-biso", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/wuc1/bi_so101_flatten-the-rag-0326-pi0-0329", "date": "2026-03-30", "verified_on": "2026-10-24", "summary": "config.json type=pi0; train_config.json dataset.repo_id=wuc1/bi_so101_flatten-the-rag-0326, whose meta/info.json robot_type=bi_so_follower (50 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/bi-so/pi0" }, { "robot": "bi-so", "policy": "pi05", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-pepijn-pi05-biso100", "ev-hub-apaszynska-pi05-biso" ] } }, "evidence": [ { "id": "ev-hub-pepijn-pi05-biso100", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/pepijn223/bimanual-so100-handover-cube_pi05", "date": "2025-10-07", "verified_on": "2026-10-24", "summary": "config.json type=pi05; train_config.json dataset.repo_id=pepijn223/bimanual-so100-handover-cube, whose meta/info.json robot_type=bi_so100_follower (25 episodes)." }, { "id": "ev-hub-apaszynska-pi05-biso", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/apaszynska/so101-cup-placement-pi05", "date": "2026-01-17", "verified_on": "2026-10-24", "summary": "config.json type=pi05; train_config.json dataset.repo_id=apaszynska/bimanual-so101-position-cup-for-coffee, whose meta/info.json robot_type=bi_so100_follower (94 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/bi-so/pi05" }, { "robot": "bi-so", "policy": "smolvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-tinkhireeva-smolvla-biso101", "ev-hub-wuc1-smolvla-biso" ] } }, "evidence": [ { "id": "ev-hub-tinkhireeva-smolvla-biso101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/tinkhireeva/bimanual-so101-stack-yellow-ring-fast-V2-smolvla", "date": "2025-10-06", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config.json dataset.repo_id=nbirukov/bimanual-so101-stack-yellow-ring-fast-V2, whose meta/info.json robot_type=bi_so101_follower (300 episodes)." }, { "id": "ev-hub-wuc1-smolvla-biso", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/wuc1/bi_so101_flatten-and-fold-the-rag-0413-model", "date": "2026-04-13", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config.json dataset.repo_id=wuc1/bi_so101_flatten-and-fold-the-rag-0331, whose meta/info.json robot_type=bi_so_follower (51 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/bi-so/smolvla" }, { "robot": "bi-so", "policy": "xvla", "state": "partial", "partial_scope": "Fine-tuning on bimanual SO data is documented upstream (so101_bimanual action mode); on-robot inference on a bimanual SO is not documented by a primary source.", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "train": { "state": "works", "evidence": [ "ev-lerobot-xvla-bimanual-train", "ev-lerobot-xvla-so101-bimanual-mode", "ev-hub-zonglin-xvla-biso100" ] } }, "evidence": [ { "id": "ev-lerobot-xvla-bimanual-train", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/xvla.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 X-VLA guide: 'Training on Bimanual Robot' example fine-tunes lerobot/xvla-base on a bimanual SO-100 dataset with --policy.action_mode=so101_bimanual.", "quote": "--dataset.repo_id=/bimanual-so100-handover-cube \\" }, { "id": "ev-lerobot-xvla-so101-bimanual-mode", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/xvla.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 X-VLA ships a dedicated so101_bimanual action mode (12-D real action = 2x(5 joints + gripper), padded to the 20-D model action).", "quote": "The `so101_bimanual` action mode handles the mismatch between model output (20D) and real robot control (12D):" }, { "id": "ev-hub-zonglin-xvla-biso100", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/zonglin11/lerobot_bi_so100_pickup_lego", "date": "2025-12-12", "verified_on": "2026-10-24", "summary": "config.json type=xvla; train_config.json dataset.repo_id=zonglin11/lerobot_bi_so100_pickup_lego, whose meta/info.json robot_type=bi_so100_follower (12 episodes)." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/bi-so/xvla" }, { "robot": "franka-fr3", "policy": "act", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Community-reported only (anonymous PhAIL benchmark under double-blind review); not documented upstream or maintainer-tested.", "Ran through the benchmark authors' own inference server (Positronic stack) with end-effector-frame actions at 15 Hz, not through a native LeRobot FR3 robot type.", "The dataset card states no per-policy success rates; of the two ACT rollouts in its sample subset, one moved all 8 items and one ran out of time at 240 s with 5 of 8 moved." ], "capabilities": { "train": { "state": "works", "notes": "ACT fine-tuned on PhAIL's 449-episode FR3 teleoperation corpus with its default open-source recipe; the checkpoint path sits under phail_unified/lerobot/.", "evidence": [ "ev-phail-card", "ev-phail-act-sample-timeout", "ev-hub-sashopepi-act" ] }, "inference_real": { "state": "works-with-caveats", "evidence": [ "ev-phail-card", "ev-phail-act-sample-timeout", "ev-phail-act-sample-success" ] } }, "evidence": [ { "id": "ev-phail-card", "tier": "community-reported", "type": "community-post", "url": "https://huggingface.co/datasets/phail-anon/phail-v1.0/blob/5bc590213c5d2615eefeb47bc23a700f8c57022a/README.md", "date": "2026-05-06", "verified_on": "2026-10-25", "summary": "PhAIL v1.0 dataset card: ~950 real-robot episodes on a Franka Research 3 with Robotiq 2F-85 (bin-to-bin pick-and-place), including ~150 ACT evaluation rollouts using ACT's default fine-tuning recipe.", "quote": "Action Chunking Transformer | `inference` | ~150 | Default fine-tuning recipe" }, { "id": "ev-phail-act-sample-timeout", "tier": "community-reported", "type": "community-post", "url": "https://huggingface.co/datasets/phail-anon/phail-v1.0/blob/5bc590213c5d2615eefeb47bc23a700f8c57022a/sample/inference/000000000000/000000000003/static.json", "date": "2026-05-06", "verified_on": "2026-10-25", "summary": "Sample ACT rollout metadata (model=act, server type act, checkpoint .../phail_unified/lerobot/270226-ee/checkpoints/220000/pretrained_model, URDF of an FR3): Batteries, 5 of 8 items moved, outcome Ran out of time after 240 s.", "quote": "\"eval.outcome\": \"Ran out of time\"" }, { "id": "ev-phail-act-sample-success", "tier": "community-reported", "type": "community-post", "url": "https://huggingface.co/datasets/phail-anon/phail-v1.0/blob/5bc590213c5d2615eefeb47bc23a700f8c57022a/sample/inference/000000000000/000000000061/static.json", "date": "2026-05-06", "verified_on": "2026-10-25", "summary": "Sample ACT rollout metadata (same checkpoint): Wooden spoons, 8 of 8 items moved in ~199 s, outcome Success.", "quote": "\"eval.outcome\": \"Success\"" }, { "id": "ev-hub-sashopepi-act", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/SashoPepi/tomato_close_box_20Ksteps", "date": "2026-03-27", "verified_on": "2026-10-25", "summary": "config.json type=act; train_config.json dataset SashoPepi/tomato_close_box (20000 steps), whose meta/info.json robot_type=fr3 (257 episodes, 8-D state, two cameras). Auto-generated card, no results." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-fr3/act" }, { "robot": "franka-fr3", "policy": "diffusion", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-sashopepi-diffusion-box", "ev-hub-sashopepi-diffusion-tomato" ] } }, "evidence": [ { "id": "ev-hub-sashopepi-diffusion-box", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/SashoPepi/lerobot_diffusion_policy_filtered", "date": "2026-03-19", "verified_on": "2026-10-25", "summary": "config.json type=diffusion; train_config.json dataset SashoPepi/franka-gello-close-box-filtered (100000 steps), whose meta/info.json robot_type=fr3 (75 episodes, 8-D state, two cameras; GELLO-teleoperated per the repo names). Auto-generated card, no results." }, { "id": "ev-hub-sashopepi-diffusion-tomato", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/SashoPepi/tomato_close_box_diffusion_3Ksteps", "date": "2026-03-26", "verified_on": "2026-10-25", "summary": "config.json type=diffusion; train_config.json dataset SashoPepi/tomato_close_box (6000 steps), whose meta/info.json robot_type=fr3 (257 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-fr3/diffusion" }, { "robot": "franka-fr3", "policy": "hil-serl", "state": "works-with-caveats", "framework": "hil-serl", "caveats": [ "Upstream names an FR3 only as one arm of its bimanual Panda + FR3 setup. It does not say whether any single-arm task ran on the FR3.", "Uses the authors' JAX codebase with ROS 1 franka_ros + serl_franka_controllers (impedance control), not LeRobot's hil-serl, whose real-robot docs do not cover Franka." ], "capabilities": { "train": { "state": "works", "notes": "Online RL with human interventions on the physical robot (dual-arm tasks such as handover, car-dashboard and timing-belt assembly).", "evidence": [ "ev-hilserl-infra", "ev-hilserl-paper-dual-arm" ] }, "inference_real": { "state": "works", "evidence": [ "ev-hilserl-infra", "ev-hilserl-paper-dual-arm" ] } }, "evidence": [ { "id": "ev-hilserl-infra", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/rail-berkeley/hil-serl/blob/c32939bccb65f3b8c43a9f9add3d322d4ab0264a/serl_robot_infra/README.md", "date": "2025-10-27", "verified_on": "2026-10-25", "summary": "serl_robot_infra README: the authors' bimanual setup pairs a Panda and an FR3, each run by an independent robot server.", "quote": "For bimmanual setup, you can run completely independent servers for each arm even if they have different firmware version (we have a Panda and a FR3) by using different catkin_ws, ROS_MASTER_URI, and flask_url." }, { "id": "ev-hilserl-paper-dual-arm", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2410.21845v3", "date": "2025-03-20", "verified_on": "2026-10-25", "summary": "HIL-SERL paper: the real-world tasks are solved with a single arm or a dual-arm setup; the dual-arm tasks are object handover, car-dashboard assembly and timing-belt assembly. The paper itself does not name the arm models.", "quote": "We solve these tasks by utilizing either a single robot arm or a dual-arm setup, together with various combinations of observations and actions." } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-fr3/hil-serl" }, { "robot": "franka-fr3", "policy": "lingbot-va", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lingbot-va", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-thu98-lingbot-fr3" ] } }, "evidence": [ { "id": "ev-hub-thu98-lingbot-fr3", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/THU98/lingbotva_franka_5task_lora_10k", "date": "2026-08-25", "verified_on": "2026-10-25", "summary": "Model card: LoRA adapter on robbyant/lingbot-va-base (official Robbyant/lingbot-va runtime, not LeRobot) trained 10k steps on 216 real-robot episodes from a dual Franka FR3 setup (16-D joint+gripper actions, chest + two wrist cameras). The training dataset is not published; adapter_config.json confirms the LoRA (rank 32). Card states no real-robot success-rate evaluation is published." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-fr3/lingbot-va" }, { "robot": "franka-fr3", "policy": "multitask-dit", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Evidence is TRI's own Large Behavior Model (DiT) stack, not LeRobot's multi_task_dit reimplementation; LeRobot has no FR3 robot type and no FR3 run through LeRobot is documented.", "TRI's platform is bimanual: two FR3 arms per station with custom compliant fingers (Schunk grippers on the evaluated platform) and 4 wrist cameras." ], "capabilities": { "train": { "state": "works", "notes": "LBMs are pretrained on a multi-source mixture (TRI's own real data, lbm_real, weighted 0.5) and finetuned per task (30k steps for real tasks).", "evidence": [ "ev-lbm-paper-finetune", "ev-lbm-paper-platform" ] }, "inference_real": { "state": "works", "notes": "Real-world evaluation on nine bimanual FR3 stations; the policy runs at 10 Hz.", "evidence": [ "ev-lbm-paper-platform", "ev-lbm-paper-hardware", "ev-lbm-paper-dit" ] } }, "evidence": [ { "id": "ev-lbm-paper-platform", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2507.05331v1", "date": "2025-07-07", "verified_on": "2026-10-25", "summary": "TRI LBM paper, Sec. 4.3.1: the real-world platform is tabletop bimanual manipulation with two Franka FR3 arms.", "quote": "We focus here on tabletop bimanual manipulation using two Franka FR3 robot arms with parallel gripper and TRI’s finray-style fingers" }, { "id": "ev-lbm-paper-hardware", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2507.05331v1", "date": "2025-07-07", "verified_on": "2026-10-25", "summary": "TRI LBM paper, Sec. 7.1: every robot station has two Franka Research 3 arms.", "quote": "All stations contain a tabletop marked with grid lines (useful for correlating object placement on the real robot stations), two Franka Research 3 arms, and two parallel jaw grippers with custom compliant fingers" }, { "id": "ev-lbm-paper-dit", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2507.05331v1", "date": "2025-07-07", "verified_on": "2026-10-25", "summary": "TRI LBM paper, Fig. 9: the LBM policy is a language-, vision- and proprioception-conditioned Diffusion Transformer run at 10 Hz on deployment.", "quote": "conditioned on language, vision and proprioception and which outputs 20-dimensional actions over 16 timesteps" }, { "id": "ev-lbm-paper-finetune", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2507.05331v1", "date": "2025-07-07", "verified_on": "2026-10-25", "summary": "TRI LBM paper: LBMs are finetuned per task on real (and simulated) data before evaluation.", "quote": "we finetune for 30k steps for real tasks and for 10k steps for simulated tasks with a global batch size of 320 and a reduced learning rate of 2e-5." }, { "id": "ev-lerobot-mtdit-lbm", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/multi_task_dit.mdx", "date": "2026-08-03", "verified_on": "2026-10-25", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 docs: LeRobot's Multitask DiT diffusion implementation is based on TRI's LBM paper, which ties this grid policy to the LBM results.", "quote": "The original diffusion implementation here is based on the work described in [TRI's LBM paper](https://arxiv.org/abs/2507.05331)" } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-fr3/multitask-dit" }, { "robot": "franka-fr3", "policy": "openvla-oft", "state": "works-with-caveats", "framework": "openvla-oft", "caveats": [ "Community-reported only (single third-party repo, ww249/Openvla-oft-for-Franka-Research-3); not documented by the OpenVLA-OFT authors or maintainer-tested.", "One task only (pick up the banana and place it into the plate), one third-person camera, Cartesian-delta actions streamed at 10 Hz through the author's own libfranka controller.", "The author reports full-task success only after capping end-effector speed and moving in small steps, and notes manually pausing near the table and above the plate before resuming the policy." ], "capabilities": { "train": { "state": "works", "notes": "Repo collects FR3 demonstrations, exports RLDS, and serves its own fine-tuned checkpoints (easy50000, hard80000) with unnorm_key franka3_banana_plate_all_success.", "evidence": [ "ev-ww249-readme", "ev-ww249-runtime" ] }, "inference_real": { "state": "works-with-caveats", "evidence": [ "ev-ww249-runtime", "ev-ww249-notes" ] } }, "evidence": [ { "id": "ev-ww249-readme", "tier": "community-reported", "type": "source-code", "url": "https://github.com/ww249/Openvla-oft-for-Franka-Research-3/blob/3584446bd7191dcb59eaf13bfa50ec1f1f51ff15/README.md", "verified_on": "2026-10-25", "summary": "Repo README (Chinese): a real Franka Research 3 data-collection repo producing single third-person-camera OpenVLA-OFT fine-tuning data; action chunk 16 x 7 at 10 Hz in the fr3_link0 frame.", "quote": "这是 Franka Research 3 真实机械臂数据采集仓库，用于采集单第三视角相机的 OpenVLA-OFT 微调数据。" }, { "id": "ev-ww249-runtime", "tier": "community-reported", "type": "source-code", "url": "https://github.com/ww249/Openvla-oft-for-Franka-Research-3/blob/3584446bd7191dcb59eaf13bfa50ec1f1f51ff15/robot_runtime/README.md", "verified_on": "2026-10-25", "summary": "robot_runtime README: OpenVLA-OFT Franka runtime queries a policy server and streams Cartesian velocity commands to the FR3 via libfranka (--execute), following the OFT deploy recipe used in training.", "quote": "The runtime follows the OpenVLA-OFT deploy recipe used during training:" }, { "id": "ev-ww249-notes", "tier": "community-reported", "type": "source-code", "url": "https://github.com/ww249/Openvla-oft-for-Franka-Research-3/blob/3584446bd7191dcb59eaf13bfa50ec1f1f51ff15/README.md", "verified_on": "2026-10-25", "summary": "README test notes dated 2026.7.19 (Chinese): with an end-effector speed cap and small per-step moves the full task succeeded; author also manually stopped near the table/above the plate before re-running the VLA.", "quote": "4. 设置机械臂末端执行器执行速度上限，每次移动小步，实现完整任务的成功" } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-fr3/openvla-oft" }, { "robot": "franka-fr3", "policy": "pi0-fast", "state": "works-with-caveats", "framework": "openpi", "caveats": [ "Community-reported only (third-party EmbodiedForge/droid-fr3, a DROID fork that replaces polymetis with the Franky SDK); not documented by Physical Intelligence or maintainer-tested.", "Documents running the DROID-fine-tuned pi0_fast_droid checkpoint (DROID 7 joint + 1 gripper action space, 15 Hz) via an openpi WebSocket server; no task success results are published." ], "capabilities": { "inference_real": { "state": "works-with-caveats", "evidence": [ "ev-droid-fr3-doc", "ev-droid-fr3-client" ] } }, "evidence": [ { "id": "ev-droid-fr3-doc", "tier": "community-reported", "type": "source-code", "url": "https://github.com/EmbodiedForge/droid-fr3/blob/a5d9d90b303b6e0f9a19c71864b7b47db6c4a60d/docs/pi0-fast-droid-inference.md", "date": "2026-05-10", "verified_on": "2026-10-25", "summary": "droid-fr3 evaluation tutorial (Chinese): run pi0_fast_droid evaluation on an already-configured FR3 with 3 calibrated ZED cameras; openpi policy server on a GPU host, client on the robot PC, 10-step chunks executed at 15 Hz.", "quote": "本文面向算法开发者，说明如何在已配置好的 FR3 机器人上运行 pi0_fast_droid 模型评测。" }, { "id": "ev-droid-fr3-client", "tier": "community-reported", "type": "source-code", "url": "https://github.com/EmbodiedForge/droid-fr3/blob/a5d9d90b303b6e0f9a19c71864b7b47db6c4a60d/scripts/server/run_policy_client.py", "date": "2026-05-10", "verified_on": "2026-10-25", "summary": "Client script: connects to an openpi server, sends DROID observations, executes the returned action chunk on the FR3 through the repo's Franky-based RobotEnv.", "quote": "PC2 policy client for pi0_fast_droid inference." }, { "id": "ev-openpi-norm-stats-fr3", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/Physical-Intelligence/openpi/blob/070759fae105d05e958efe78fe6fb08631f23da5/docs/norm_stats.md", "date": "2025-06-20", "verified_on": "2026-10-25", "summary": "Supporting only (not deployment evidence): openpi publishes pi0_base and pi0_fast_base pre-training normalization statistics for an FR3 + Robotiq 2F-85 setup (asset 'franka', separate from the DROID asset used above).", "quote": "Franka Emika (non-DROID) | Franka FR3 arm with Robotiq 2F-85 gripper | franka" } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-fr3/pi0-fast" }, { "robot": "franka-fr3", "policy": "pi05", "state": "works-with-caveats", "framework": "openpi", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "The physical deployment is through OpenPI (pi05_droid config, 15 Hz: seven FR3 joint velocities plus gripper closedness) with the authors' guarded client; the LeRobot evidence here is a training checkpoint only.", "No aggregate success metrics are published; the authors withhold them until their v3_12000 validation ledger is reviewed, and the only public rollout evidence is a linked demo video.", "The checkpoint is tied to one cube pick-and-place workcell (two RealSense cameras, its calibration and safety guards) and the authors say not to use it on another workcell without commissioning." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-bekhzod-pi05", "ev-card-msbutt1-pi05-fr3" ] }, "inference_real": { "state": "works-with-caveats", "notes": "Guarded live evaluation on a Franka Research 3 via OpenPI serve_policy; no success rate published.", "evidence": [ "ev-card-msbutt1-pi05-fr3", "ev-repo-msbutt1-fr3" ] } }, "evidence": [ { "id": "ev-hub-bekhzod-pi05", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Bekhzod/Pi05_fr3_pickplace_fixed_goal_robustness_recovery_data_model", "date": "2026-07-17", "verified_on": "2026-10-24", "summary": "config.json type=pi05 (pretrained_path lerobot/pi05_base); train_config dataset Bekhzod/fr3_pickplace_fixed_goal_robustness_recovery_merged_data (350 episodes, front + top cameras), whose meta/info.json robot_type=FR3WMS. Auto-generated card, no rollout results." }, { "id": "ev-card-msbutt1-pi05-fr3", "tier": "community-reported", "type": "model-card", "url": "https://huggingface.co/msbutt1/pi0.5-fr3-v3-pick-place/blob/46173e13f41e5dc8dba8e1ee29f4fbf71a95c236/README.md", "date": "2026-07-22", "verified_on": "2026-10-25", "summary": "OpenPI checkpoint card: pi0.5-DROID fine-tuned on a Franka Research 3 (FR3, not Panda) cube pick-and-place dataset; deployment uses a local FR3 reference streamer, workspace/joint guards, grasp verification and an E-stop; aggregate success metrics intentionally not reported.", "quote": "This checkpoint is a pi0.5-DROID policy fine-tuned for real-world\npick-and-place with a Franka Research 3 (FR3) robot." }, { "id": "ev-repo-msbutt1-fr3", "tier": "community-reported", "type": "source-code", "url": "https://github.com/msbutt1/franka-fr3-robot-learning/blob/14760a4f6738e393ea9815d539b8ac05e4ad88a2/README.md", "date": "2026-10-02", "verified_on": "2026-10-25", "summary": "The authors' FR3 repo documents serving this checkpoint with OpenPI and running it from a libfranka/FCI robot workstation, and links a Live Policy Evaluations wiki page whose 'Pick and Place' entry is a demo video (title read via oEmbed; footage not independently reviewed). It also warns the bracket/bolt collection config uses a different absolute-joint action schema.", "quote": "The released cube policy and the dated bracket evaluation use pi0.5-DROID\njoint-velocity actions: seven FR3 joint velocities plus gripper closedness at\n15 Hz." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-fr3/pi05" }, { "robot": "franka-fr3", "policy": "smolvla", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Community-reported only (CRISP paper authors); not documented upstream or maintainer-tested.", "Runs through the third-party CRISP ROS 2 controllers and crisp_gym interface; LeRobot v0.6.1 has no native FR3 robot type.", "SmolVLA ran at approximately 10 Hz and outputs end-effector target poses tracked by CRISP's Cartesian impedance controllers; the paper reports no task success rates." ], "capabilities": { "train": { "state": "works", "notes": "Data recorded on the FR3 in LeRobotDataset format and trained with LeRobot's training scripts.", "evidence": [ "ev-lerobot-issue-1567-crisp", "ev-crisp-paper-deploy", "ev-phail-card-smolvla" ] }, "inference_real": { "state": "works-with-caveats", "evidence": [ "ev-crisp-paper-deploy", "ev-crisp-paper-fr3", "ev-phail-card-smolvla" ] } }, "evidence": [ { "id": "ev-crisp-paper-deploy", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2509.06819v2", "date": "2026-02-06", "verified_on": "2026-10-25", "summary": "CRISP paper, Fig. 5: Diffusion Policy and SmolVLA trained on CRISP_GYM-recorded data are deployed on hardware; SmolVLA at ~10 Hz.", "quote": "Using the same recorded data in the CRISP_GYM interface, different learned policies can be successfully deployed in hardware. Left: Diffusion Policy [4] operating at approximately 30 Hz. Right: SmolVLA [27] operating at approximately 10 Hz." }, { "id": "ev-crisp-paper-fr3", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2509.06819v2", "date": "2026-02-06", "verified_on": "2026-10-25", "summary": "CRISP paper, Sec. IV: the only hardware platform is the Franka Robotics FR3 (other arms are simulation only).", "quote": "The controllers have been validated on hardware using the Franka Robotics FR3 and in simulation with the Kuka IIWA14 and Kinova Gen3, as shown in Figure 3." }, { "id": "ev-lerobot-issue-1567-crisp", "tier": "community-reported", "type": "github-issue", "url": "https://github.com/huggingface/lerobot/issues/1567#issuecomment-3188267089", "date": "2025-08-14", "verified_on": "2026-10-25", "summary": "CRISP author (LeRobot contributor) reports training VLAs including SmolVLA with LeRobot on a Franka Robotics FR3, recording LeRobotDataset data and deploying with their own script.", "quote": "At our lab we have been training VLAs (also tried smolVLA) using lerobot and a Franka Robotics FR3." }, { "id": "ev-phail-card-smolvla", "tier": "community-reported", "type": "community-post", "url": "https://huggingface.co/datasets/phail-anon/phail-v1.0/blob/5bc590213c5d2615eefeb47bc23a700f8c57022a/README.md", "date": "2026-05-06", "verified_on": "2026-10-25", "summary": "PhAIL v1.0 (anonymous benchmark under double-blind review): ~115 SmolVLA evaluation rollouts on a Franka Research 3 with Robotiq 2F-85 (bin-to-bin pick-and-place), SmolVLA fine-tuned on the 449-episode FR3 corpus with its default recipe; no per-policy success rates on the card.", "quote": "Hugging Face SmolVLA | `inference` | ~115 | Default fine-tuning recipe" } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-fr3/smolvla" }, { "robot": "franka-fr3", "policy": "vla-jepa", "state": "works-with-caveats", "framework": "vla-jepa", "caveats": [ "Evidence is the authors' own lab evaluation (100 demonstrations, three pick-and-place tasks); the upstream repo ships training and simulation-evaluation scripts but no FR3 deployment code or FR3 checkpoint." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-vlajepa-paper" ] }, "inference_real": { "state": "works", "evidence": [ "ev-vlajepa-paper" ] } }, "evidence": [ { "id": "ev-vlajepa-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2602.10098", "date": "2026-02-10", "verified_on": "2026-10-24", "summary": "The VLA-JEPA paper's real-world experiments fine-tune and evaluate VLA-JEPA on a Franka Research 3 with a Robotiq 2F-85 gripper and three RealSense D435 cameras (two third-person, one wrist).", "quote": "For real-world experiments, we designed table-top manipulation tasks using a Franka Research 3 arm equipped with a Robotiq 2F-85 gripper. We collected 100 human demonstration trajectories for training, which include 3 picking and placing tasks." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-fr3/vla-jepa" }, { "robot": "franka-fr3", "policy": "xvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "caveats": [ "The FR3 variant comes from the dataset name only; its meta/info.json robot_type is the generic \"franka\"." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-binsabit-xvla" ] } }, "evidence": [ { "id": "ev-hub-binsabit-xvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/binsabit/xvla-franka-fr3", "date": "2026-04-07", "verified_on": "2026-10-24", "summary": "config.json type=xvla (pretrained_path lerobot/xvla-base); train_config dataset binsabit/franka-fr3-manipulation (101 episodes, stationary + wrist cameras), whose meta/info.json robot_type=franka. Auto-generated card; repo includes two .mp4 files that are not described as policy rollouts." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-fr3/xvla" }, { "robot": "franka-panda", "policy": "act", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "The real-robot evidence uses the LeFranX fork (Panda/FER + RobotEra XHand, 19-D action space), not an unmodified LeRobot install. LeFranX has no Franka Hand gripper interface." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-levr-paper", "ev-hub-tinkybala-act", "ev-hub-ases-act" ] }, "inference_real": { "state": "works", "notes": "LeVR: three tasks, 10 rollouts per task-policy pair, remote inference via LeRobot's gRPC policy server.", "evidence": [ "ev-levr-paper", "ev-levr-paper-hw" ] } }, "evidence": [ { "id": "ev-levr-paper", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2509.14349v1", "date": "2025-09-17", "verified_on": "2026-10-24", "summary": "LeVR paper: LeRobot ACT is fine-tuned on 100 LeFranX demos per task and rolled out on the robot (10 rollouts per task, 3 tasks).", "quote": "Using the LeRobot framework, we directly finetuned two state-of-the-art visuomotor policies, ACT and DP, on our LeFranX-collected dataset." }, { "id": "ev-levr-paper-hw", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2509.14349v1", "date": "2025-09-17", "verified_on": "2026-10-24", "summary": "LeVR hardware: Franka FER with an XHand, three RealSense D435 cameras at 320x240/30 fps, inference on an RTX 6000 Ada server.", "quote": "Experiments were conducted using a Franka FER robot equipped with a RobotEra XHand dexterous hand." }, { "id": "ev-hub-tinkybala-act", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/TInkybala/Real_Panda_PickCubeVLA0_act_20260402_143912", "date": "2026-04-02", "verified_on": "2026-10-24", "summary": "config.json type=act. train_config dataset TInkybala/PickCubeVLA0, whose meta/info.json robot_type=panda_ros_cartesian (148 episodes, panda_joint1-7 state, Cartesian-velocity actions). Auto-generated card, no rollout results. The Panda identity comes from the repo/robot_type naming." }, { "id": "ev-hub-ases-act", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/ases200q2/Real_Panda_CartesianVel_SpaceMouse_policy", "verified_on": "2026-10-24", "summary": "config.json type=act. train_config dataset ases200q2/Real_Panda_CartesianVel_SpaceMouse, whose meta/info.json robot_type=panda_ros_cartesian (25 episodes, one camera)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/act" }, { "robot": "franka-panda", "policy": "diffusion", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "The LeRobot route is community-reported only: the LeVR/LeFranX authors fine-tuned LeRobot's Diffusion Policy on a Panda (FER) with a RobotEra XHand, using the LeFranX fork, which has no Franka Hand gripper support.", "The Diffusion Policy authors' own Panda experiments (sauce pouring/spreading, bimanual tasks) ran on their own stack. The public real-stanford/diffusion_policy real-robot code targets UR5 only." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-levr-paper", "ev-dp-paper-panda", "ev-droid-paper-dp" ] }, "inference_real": { "state": "works", "notes": "Upstream: Diffusion Policy paper (Panda). Also the DROID paper (robomimic-based droid_policy_learning) and OpenVLA's from-scratch baseline. Via LeRobot: LeVR, 10 rollouts per task.", "evidence": [ "ev-dp-paper-panda", "ev-droid-paper-dp", "ev-openvla-paper-dp-baseline", "ev-levr-paper" ] } }, "evidence": [ { "id": "ev-levr-paper", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2509.14349v1", "date": "2025-09-17", "verified_on": "2026-10-24", "summary": "LeVR paper: ACT and Diffusion Policy as hosted by LeRobot are fine-tuned on 100 demos per task, collected with LeFranX on a Franka FER + XHand, and rolled out on the robot (10 rollouts per task, 3 tasks).", "quote": "Using the LeRobot framework, we directly finetuned two state-of-the-art visuomotor policies, ACT and DP, on our LeFranX-collected dataset." }, { "id": "ev-dp-paper-panda", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2303.04137v5", "date": "2024-03-14", "verified_on": "2026-10-24", "summary": "Diffusion Policy paper (v5): real-world 6-DoF sauce pouring and periodic spreading tasks on a Franka Panda, plus bimanual tasks teleoperated onto Franka Panda arms.", "quote": "Our Franka Panda setup and tasks are shown in Fig 10." }, { "id": "ev-droid-paper-dp", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2403.12945v2", "date": "2025-04-22", "verified_on": "2026-10-24", "summary": "DROID paper: all policy-learning experiments train diffusion policies (robomimic-based, released as droid_policy_learning), evaluated on 6 real tasks on the Panda-based DROID hardware.", "quote": "To this end, we use diffusion policies [7, 16, 21, 38, 54], which leverage denoising diffusion models for action prediction" }, { "id": "ev-openvla-paper-dp-baseline", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2406.09246v3", "date": "2024-09-05", "verified_on": "2026-10-24", "summary": "OpenVLA paper trains Diffusion Policy from scratch on seven Franka Emika Panda tasks as a baseline. It is strong on narrow single-instruction tasks and weaker on multi-instruction ones.", "quote": "Diffusion Policy exhibits strong performance on narrow single-instruction tasks, while Octo and OpenVLA perform better on diverse fine-tuning tasks involving multiple instructions and distractor objects." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/diffusion" }, { "robot": "franka-panda", "policy": "eo1", "state": "works", "framework": "eo1", "capabilities": { "train": { "state": "works", "evidence": [ "ev-eo1-paper-panda" ] }, "inference_real": { "state": "works", "notes": "Seven Franka Panda pick-and-place tasks plus reasoning-control and open-world tasks. The upstream client controls the Panda through Deoxys (NUC with a real-time kernel + GPU workstation) with RealSense cameras.", "evidence": [ "ev-eo1-paper-panda", "ev-eo1-repo-franka" ] } }, "evidence": [ { "id": "ev-eo1-paper-panda", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2508.21112v5", "date": "2026-02-25", "verified_on": "2026-10-24", "summary": "EO-1 paper (v5), Sec. 4.5: EO-1 is evaluated in the real world on a Franka Panda (7 pick-and-place tasks) among other robots.", "quote": "We evaluate its performance on both short-horizon and long-horizon tasks using different robots, including Franka Panda, WidowX 250 S, Agibot G-1, and Lerobot SO100." }, { "id": "ev-eo1-repo-franka", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/EO-Robotics/EO1/blob/451c363d7dff682700789afae0b9b17c004e8f8c/experiments/7_franka/README.md", "date": "2025-11-12", "verified_on": "2026-10-24", "summary": "EO1 repo ships a Franka Emika Panda real-robot client (eval_franka.py, deoxys_control submodule).", "quote": "This directory contains the implementation for controlling Franka Emika Panda robots using the EO-1 model." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/eo1" }, { "robot": "franka-panda", "policy": "evo1", "state": "partial", "partial_scope": "Simulation only: LIBERO (simulated Franka Panda); no real-robot training or inference documented.", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "inference_sim": { "state": "works", "notes": "LeRobot v0.6.1 reports 96.65% overall on LIBERO (2,000 episodes) for its reproduced Stage-2 checkpoint; the Evo-1 paper reports 94.8%.", "evidence": [ "ev-lerobot-evo1-libero", "ev-evo1-paper-libero", "ev-libero-panda-src" ] } }, "evidence": [ { "id": "ev-lerobot-evo1-libero", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/evo1.mdx", "date": "2026-08-03", "verified_on": "2026-10-25", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 Evo-1 guide: two-stage training on lerobot/libero and a LIBERO evaluation of the released Stage-2 checkpoint (zuoxingdong/evo1_libero).", "quote": "| **Overall** | **1,933** | **2,000** | **96.65%** |" }, { "id": "ev-evo1-paper-libero", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2511.04555v2", "verified_on": "2026-10-25", "summary": "Evo-1 paper (v2): LIBERO is one of its three simulation benchmarks (with Meta-World and RoboTwin).", "quote": "and reaches 94.8% on LIBERO, demonstrating its adaptability in both single-arm and dual-arm manipulation tasks." }, { "id": "ev-libero-panda-src", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/Lifelong-Robot-Learning/LIBERO/blob/8f1084e3132a39270c3a13ebe37270a43ece2a01/libero/libero/envs/robots/mounted_panda.py", "date": "2025-03-15", "verified_on": "2026-10-25", "summary": "LIBERO's tabletop tasks instantiate a Mounted model (libero_tabletop_manipulation.py); its MountedPanda robot class is the Franka Panda.", "quote": "Panda is a sensitive single-arm robot designed by Franka." } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/evo1" }, { "robot": "franka-panda", "policy": "groot-n1-7", "state": "works-with-caveats", "framework": "isaac-groot", "caveats": [ "NVIDIA documents the DROID deployment path (inference server + DROID robot control script) but publishes no on-robot DROID success rates for N1.7. The README reports only offline MSE on demo data.", "Targets the DROID platform (ZED cameras, DROID control package), not an arbitrary Panda setup. The DROID software accepts either a Panda or an FR3." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-groot-droid-example" ] }, "inference_real": { "state": "works-with-caveats", "notes": "Deployment path documented; no published real-robot results.", "evidence": [ "ev-groot-droid-example", "ev-groot-droid-server", "ev-droid-paper" ] } }, "evidence": [ { "id": "ev-groot-droid-example", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/NVIDIA/Isaac-GR00T/blob/23ace64f17aa5015259b8609d371eb61a357c776/examples/DROID/README.md", "date": "2026-04-18", "verified_on": "2026-10-24", "summary": "Isaac-GR00T n1.7-release DROID example: zero-shot via the OXE_DROID embodiment tag or the nvidia/GR00T-N1.7-DROID fine-tune, fine-tuning commands, and a robot control script (main_gr00t.py) for the DROID platform.", "quote": "The N1.7 base model supports DROID inference out of the box via the `OXE_DROID_RELATIVE_EEF_RELATIVE_JOINT` pretrain tag." }, { "id": "ev-groot-droid-server", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/NVIDIA/Isaac-GR00T/blob/23ace64f17aa5015259b8609d371eb61a357c776/examples/DROID/README.md", "date": "2026-04-18", "verified_on": "2026-10-24", "summary": "The same README documents an inference server for real-world deployment and a client that runs on the DROID control laptop.", "quote": "Inference Server (for real-world deployment)" }, { "id": "ev-droid-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2403.12945v2", "date": "2025-04-22", "verified_on": "2026-10-24", "summary": "The DROID platform and dataset are built on the Franka Emika Panda.", "quote": "We chose the Franka Emika Panda 7 DoF robot arm as the base of our setup since it is widely adopted in the robot research community, reliable, relatively affordable and was available at most participating institutions." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/groot-n1-7" }, { "robot": "franka-panda", "policy": "hil-serl", "state": "works-with-caveats", "framework": "hil-serl", "caveats": [ "Upstream names a Panda only as one arm of its bimanual Panda + FR3 setup. It does not say which single-arm tasks ran on which Franka model.", "Uses the authors' JAX codebase with ROS 1 franka_ros + serl_franka_controllers (impedance control), not LeRobot's hil-serl, whose real-robot docs do not cover Franka." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hilserl-readme", "ev-hilserl-infra" ] }, "inference_real": { "state": "works", "evidence": [ "ev-hilserl-readme", "ev-hilserl-infra" ] } }, "evidence": [ { "id": "ev-hilserl-readme", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/rail-berkeley/hil-serl/blob/c32939bccb65f3b8c43a9f9add3d322d4ab0264a/README.md", "date": "2025-10-27", "verified_on": "2026-10-24", "summary": "HIL-SERL authors' repo provides a Franka walkthrough (RAM insertion, USB pick-up and insertion, object handover, egg flip).", "quote": "We provide a step-by-step guide to run RL policies with HIL-SERL on a Franka robot." }, { "id": "ev-hilserl-infra", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/rail-berkeley/hil-serl/blob/c32939bccb65f3b8c43a9f9add3d322d4ab0264a/serl_robot_infra/README.md", "date": "2025-10-27", "verified_on": "2026-10-24", "summary": "serl_robot_infra README: the authors' bimanual setup pairs a Panda and an FR3, each run by an independent robot server.", "quote": "For bimmanual setup, you can run completely independent servers for each arm even if they have different firmware version (we have a Panda and a FR3) by using different catkin_ws, ROS_MASTER_URI, and flask_url." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/hil-serl" }, { "robot": "franka-panda", "policy": "molmoact2", "state": "works-with-caveats", "framework": "molmoact2", "caveats": [ "Evidence is for the DROID setup (Franka, external + wrist camera, 15-step chunks at DROID's 15 Hz). The paper says 'DROID setup' without naming the arm model. The DROID platform is Panda-based, but its software also accepts an FR3.", "The upstream repo ships the MolmoAct2-DROID inference server (examples/droid/host_server_droid.py) but not the robot-side client." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-molmoact2-paper-droid-ft" ] }, "inference_real": { "state": "works", "notes": "Zero-shot real-world DROID evaluation: five kitchen tasks x three spatial variants, 87.1% (vs. MolmoBot and π0 baselines).", "evidence": [ "ev-molmoact2-paper-droid-real", "ev-molmoact2-droid-server", "ev-lerobot-molmoact2-doc", "ev-droid-paper" ] } }, "evidence": [ { "id": "ev-molmoact2-paper-droid-real", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2605.02881v2", "date": "2026-05-08", "verified_on": "2026-10-24", "summary": "MolmoAct2 paper, Sec. 6.2 / App. C.2: MolmoAct2-DROID is evaluated zero-shot on a real DROID setup in novel scenes, reaching 87.1%.", "quote": "In real-world evaluation on the DROID setup, which is substantially more out-of-distribution given the random camera initialization and entirely novel scenes and objects relative to the DROID dataset (Khazatsky et al., 2024), MolmoAct2 again attains the highest performance, reaching 87.1%" }, { "id": "ev-molmoact2-paper-droid-ft", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2605.02881v2", "date": "2026-05-08", "verified_on": "2026-10-24", "summary": "MolmoAct2 paper: the MolmoAct2-DROID checkpoint is fine-tuned on a quality-filtered DROID mixture.", "quote": "The MolmoAct2-DROID checkpoint is fine-tuned on the filtered DROID mixture." }, { "id": "ev-molmoact2-droid-server", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/allenai/molmoact2/blob/6070080a20321b4f498ab30f28e1d09ac465edb7/examples/droid/host_server_droid.py", "date": "2026-10-02", "verified_on": "2026-10-24", "summary": "Upstream MolmoAct2-DROID HTTP inference server (allenai/MolmoAct2-DROID, norm tag franka_droid). It takes an external + wrist image and an 8-D joint+gripper state.", "quote": "MolmoAct2-DROID inference server." }, { "id": "ev-lerobot-molmoact2-doc", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/molmoact2.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 lists a LeRobot-format MolmoAct2-DROID checkpoint for lerobot-rollout. LeRobot has no Panda robot type, so that route needs a third-party robot integration.", "quote": "| MolmoAct2-DROID-LeRobot | `cam0`, `cam1` | External + wrist |" }, { "id": "ev-droid-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2403.12945v2", "date": "2025-04-22", "verified_on": "2026-10-24", "summary": "The DROID platform and dataset are built on the Franka Emika Panda.", "quote": "We chose the Franka Emika Panda 7 DoF robot arm as the base of our setup since it is widely adopted in the robot research community, reliable, relatively affordable and was available at most participating institutions." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/molmoact2" }, { "robot": "franka-panda", "policy": "openvla-oft", "state": "partial", "partial_scope": "Simulation only: LIBERO; no real-robot training or inference documented.", "framework": "openvla-oft", "capabilities": { "inference_sim": { "state": "works", "notes": "Fine-tuned and evaluated on the four LIBERO suites (Spatial, Object, Goal, Long); upstream releases LIBERO-finetuned checkpoints.", "evidence": [ "ev-oft-paper-libero", "ev-oft-readme-libero" ] } }, "evidence": [ { "id": "ev-oft-paper-libero", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2502.19645v2", "date": "2025-04-28", "verified_on": "2026-10-25", "summary": "OpenVLA-OFT paper, Sec. V-A: LIBERO, the simulation benchmark used, features a simulated Franka Emika Panda arm.", "quote": "We evaluate on the LIBERO simulation benchmark [26], which features a Franka Emika Panda arm in simulation with demonstrations containing camera images, robot state, task annotations, and delta end-effector pose actions." }, { "id": "ev-oft-readme-libero", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/moojink/openvla-oft/blob/e4287e94541f459edc4feabc4e181f537cd569a8/README.md", "verified_on": "2026-10-25", "summary": "OpenVLA-OFT README quick start loads the LIBERO-Spatial fine-tuned checkpoint and lists the GPU needed for LIBERO sim tasks.", "quote": "1 GPU with ~16 GB VRAM for LIBERO sim benchmark tasks" } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/openvla-oft" }, { "robot": "franka-panda", "policy": "openvla", "state": "works-with-caveats", "framework": "openvla", "caveats": [ "Panda results come from fine-tuning (10-150 demonstrations per task), not out-of-the-box use. The OpenVLA repo ships no Franka controller or deployment script.", "OpenVLA takes a single image and predicts single-step actions. The Franka setups ran 5 Hz (Franka-Tabletop) and 15 Hz (Franka-DROID) non-blocking controllers." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-openvla-paper-franka" ] }, "inference_real": { "state": "works", "evidence": [ "ev-openvla-paper-franka", "ev-openvla-paper-franka-tasks" ] } }, "evidence": [ { "id": "ev-openvla-paper-franka", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2406.09246v3", "date": "2024-09-05", "verified_on": "2026-10-24", "summary": "OpenVLA paper, Sec. 5.2: OpenVLA is fully fine-tuned and evaluated on a table-mounted Franka Emika Panda (Franka-Tabletop) and on the DROID Franka setup.", "quote": "We test OpenVLA in two setups: Franka-Tabletop, a stationary, table-mounted Franka Emika Panda 7-DoF robot arm; and Franka-DROID, the Franka robot arm setup from the recently released DROID dataset [11]," }, { "id": "ev-openvla-paper-franka-tasks", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2406.09246v3", "date": "2024-09-05", "verified_on": "2026-10-24", "summary": "OpenVLA paper, Appendix B.3: six Franka-Tabletop (Panda) tasks and one Franka-DROID task, 129 rollouts per approach.", "quote": "We collect 10–150 demonstrations of each of seven tasks. The first six tasks correspond to a robot setup which we denote as “Franka-Tabletop” (Franka Emika Panda robot mounted on top of a table)," } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/openvla" }, { "robot": "franka-panda", "policy": "pi0-fast", "state": "works-with-caveats", "framework": "openpi", "caveats": [ "Documented for the DROID platform (Panda, two ZED 2 exterior cameras plus a ZED Mini wrist camera, DROID control laptop/NUC), not for an arbitrary Panda setup. openpi warns that its expert checkpoints may not generalize to other setups.", "Inference runs on a remote GPU policy server (~RTX 4090). openpi notes that 0.5-1 s latency per action chunk is normal." ], "capabilities": { "train": { "state": "works", "notes": "openpi documents training generalist policies on the full DROID dataset (approximate open-source pipeline for pi0-FAST-DROID).", "evidence": [ "ev-openpi-readme-fast-droid", "ev-fast-paper-droid" ] }, "inference_real": { "state": "works", "evidence": [ "ev-fast-paper-droid", "ev-openpi-readme-fast-droid", "ev-openpi-droid-example", "ev-droid-paper" ] } }, "evidence": [ { "id": "ev-fast-paper-droid", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2501.09747", "date": "2025-01-16", "verified_on": "2026-10-24", "summary": "π0-FAST paper: a FAST policy trained on DROID is evaluated zero-shot on real DROID tabletop tasks in unseen environments across three university campuses.", "quote": "Notably, FAST tokenization enables the first successful training of a strong generalist policy on the DROID dataset [38], which can be evaluated zero-shot in unseen environments, without fine-tuning, by simply prompting it in natural language." }, { "id": "ev-openpi-readme-fast-droid", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/README.md", "date": "2026-08-24", "verified_on": "2026-10-24", "summary": "openpi lists the pi0_fast_droid 'Inference' checkpoint for 0-shot use on the DROID robot platform.", "quote": "$\\pi_0$-FAST model fine-tuned on the [DROID dataset](https://droid-dataset.github.io/): can perform a wide range of simple table-top manipulation tasks 0-shot in new scenes on the DROID robot platform" }, { "id": "ev-openpi-droid-example", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/examples/droid/README.md", "date": "2026-08-24", "verified_on": "2026-10-24", "summary": "openpi DROID example: serve pi0_fast_droid and run the DROID robot client (main.py) on the DROID control laptop.", "quote": "uv run scripts/serve_policy.py policy:checkpoint --policy.config=pi0_fast_droid --policy.dir=gs://openpi-assets/checkpoints/pi0_fast_droid" }, { "id": "ev-droid-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2403.12945v2", "date": "2025-04-22", "verified_on": "2026-10-24", "summary": "The DROID platform and dataset are built on the Franka Emika Panda. (The DROID control software also accepts an FR3, so the same openpi route plausibly covers FR3-based DROID rigs.)", "quote": "We chose the Franka Emika Panda 7 DoF robot arm as the base of our setup since it is widely adopted in the robot research community, reliable, relatively affordable and was available at most participating institutions." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/pi0-fast" }, { "robot": "franka-panda", "policy": "pi0", "state": "works-with-caveats", "framework": "openpi", "caveats": [ "openpi lists pi0_droid as a fine-tuning starting point and says it may not follow language commands as well as π0-FAST-DROID. On real DROID tasks PI reports a lower score for π0 (0.49) than for its knowledge-insulated model (0.55).", "Documented for the DROID platform (Panda + ZED cameras + DROID control stack) through a remote policy server, not for an arbitrary Panda setup." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-openpi-readme-pi0-droid" ] }, "inference_real": { "state": "works", "evidence": [ "ev-ki-paper-droid", "ev-openpi-droid-example", "ev-droid-paper" ] } }, "evidence": [ { "id": "ev-ki-paper-droid", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2505.23705", "date": "2025-05-29", "verified_on": "2026-10-24", "summary": "Physical Intelligence's knowledge-insulation paper evaluates π0 (and its own method) on the real-world DROID benchmark in unseen scenes. π0 scored 0.49 ± 0.09 task progress.", "quote": "We also evaluate our generalist on the open source benchmark DROID [23] for the same set of tabletop manipulation tasks as in [37]. Our method received a score of 0.55 ± 0.09, π0 received 0.49 ± 0.09" }, { "id": "ev-openpi-readme-pi0-droid", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/README.md", "date": "2026-08-24", "verified_on": "2026-10-24", "summary": "openpi lists the pi0_droid checkpoint (usage: Fine-Tuning).", "quote": "$\\pi_0$ model fine-tuned on the [DROID dataset](https://droid-dataset.github.io/): faster inference than $\\pi_0$-FAST-DROID, but may not follow language commands as well" }, { "id": "ev-openpi-droid-example", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/examples/droid/README.md", "date": "2026-08-24", "verified_on": "2026-10-24", "summary": "openpi DROID example gives a server config for pi0_droid, then the same DROID robot client as for π0.5.", "quote": "uv run scripts/serve_policy.py policy:checkpoint --policy.config=pi0_droid --policy.dir=gs://openpi-assets/checkpoints/pi0_droid" }, { "id": "ev-droid-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2403.12945v2", "date": "2025-04-22", "verified_on": "2026-10-24", "summary": "The DROID platform and dataset are built on the Franka Emika Panda.", "quote": "We chose the Franka Emika Panda 7 DoF robot arm as the base of our setup since it is widely adopted in the robot research community, reliable, relatively affordable and was available at most participating institutions." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/pi0" }, { "robot": "franka-panda", "policy": "pi05", "state": "works-with-caveats", "framework": "openpi", "caveats": [ "Documented for the DROID platform (Panda, ZED 2 exterior + ZED Mini wrist cameras, DROID control laptop/NUC), not for an arbitrary Panda setup. The policy uses one exterior camera plus the wrist camera.", "Inference runs on a remote GPU policy server (~RTX 4090). openpi notes that 0.5-1 s latency per action chunk is normal and that the policy fails on more complex manipulation tasks." ], "capabilities": { "train": { "state": "works", "notes": "openpi documents fine-tuning π0.5 on a custom DROID dataset.", "evidence": [ "ev-openpi-readme-pi05-droid" ] }, "inference_real": { "state": "works", "evidence": [ "ev-openpi-droid-example", "ev-ki-paper-droid", "ev-droid-paper" ] } }, "evidence": [ { "id": "ev-openpi-droid-example", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/examples/droid/README.md", "date": "2026-08-24", "verified_on": "2026-10-24", "summary": "openpi's DROID example runs pi05_droid on the DROID robot platform via a policy server and the DROID robot client, and calls it PI's strongest generalist DROID policy on RoboArena.", "quote": "This example shows how to run the fine-tuned $\\pi_{0.5}$-DROID model on the [DROID robot platform](https://github.com/droid-dataset/droid)." }, { "id": "ev-openpi-readme-pi05-droid", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/README.md", "date": "2026-08-24", "verified_on": "2026-10-24", "summary": "openpi lists pi05_droid (usage: Inference / Fine-Tuning), a π0.5 fine-tuned on DROID with knowledge insulation.", "quote": "$\\pi_{0.5}$ model fine-tuned on the [DROID dataset](https://droid-dataset.github.io/) with [knowledge insulation](https://www.physicalintelligence.company/research/knowledge_insulation): fast inference and good language-following" }, { "id": "ev-ki-paper-droid", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2505.23705", "date": "2025-05-29", "verified_on": "2026-10-24", "summary": "PI's knowledge-insulation paper (the recipe behind pi05_droid per openpi) evaluates a DROID specialist on the real-world DROID benchmark in unseen scenes: 0.55 ± 0.09 task progress.", "quote": "We evaluate our method on the simulated LIBERO [30] benchmark, as well as the real world DROID [23] benchmark." }, { "id": "ev-droid-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2403.12945v2", "date": "2025-04-22", "verified_on": "2026-10-24", "summary": "The DROID platform and dataset are built on the Franka Emika Panda.", "quote": "We chose the Franka Emika Panda 7 DoF robot arm as the base of our setup since it is widely adopted in the robot research community, reliable, relatively affordable and was available at most participating institutions." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/pi05" }, { "robot": "franka-panda", "policy": "smolvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "caveats": [ "The checkpoints' Panda identity comes from repo and robot_type naming (panda_ros_cartesian, panda_joint* states). The cards are auto-generated with no hardware description." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-tinkybala-smolvla-screwdriver", "ev-hub-tinkybala-smolvla-cube" ] } }, "evidence": [ { "id": "ev-hub-tinkybala-smolvla-screwdriver", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/TInkybala/Real_Panda_PickScrewdriver_smolvla_20260304_151600", "date": "2026-03-04", "verified_on": "2026-10-24", "summary": "config.json type=smolvla (base lerobot/smolvla_base). train_config dataset TInkybala/Real_Panda_CartesianVel_Screwdriver_2, whose meta/info.json robot_type=panda_ros_cartesian (75 episodes, one camera)." }, { "id": "ev-hub-tinkybala-smolvla-cube", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/TInkybala/Real_Panda_Fixed_Cube", "verified_on": "2026-10-24", "summary": "config.json type=smolvla. train_config dataset TInkybala/Real_Panda_FixedCube, whose meta/info.json robot_type=panda_ros_cartesian (24 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/smolvla" }, { "robot": "franka-panda", "policy": "vla-jepa", "state": "partial", "partial_scope": "Simulation only: LIBERO (simulated Franka Emika Panda); no real-robot training or inference on a Panda is documented. The paper's real-robot experiments use a Franka Research 3 (see the franka-fr3 cell).", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "inference_sim": { "state": "works", "notes": "LIBERO (four suites) in the paper; LeRobot v0.6.1 documents fine-tuning on HuggingFaceVLA/libero and evaluating lerobot/VLA-JEPA-LIBERO with lerobot-eval --env.type=libero.", "evidence": [ "ev-vlajepa-paper-libero", "ev-lerobot-vlajepa-libero" ] } }, "evidence": [ { "id": "ev-vlajepa-paper-libero", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2602.10098", "date": "2026-02-10", "verified_on": "2026-10-25", "summary": "VLA-JEPA paper, Sec. 4.2: VLA-JEPA is evaluated on the LIBERO simulation benchmark, which the paper states uses the Franka Emika Panda arm.", "quote": "The LIBERO benchmark employs the Franka Emika Panda arm and comprises four task suites designed to facilitate research on lifelong learning in robotic manipulation." }, { "id": "ev-lerobot-vlajepa-libero", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/vla_jepa.mdx", "date": "2026-08-03", "verified_on": "2026-10-25", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 VLA-JEPA doc: 'Reproducing the LIBERO results' section with a lerobot-train command on HuggingFaceVLA/libero and lerobot-eval of lerobot/VLA-JEPA-LIBERO on the LIBERO env.", "quote": "starts the training from the Pretrain checkpoint, trains for 30k steps on the LIBERO dataset." } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/vla-jepa" }, { "robot": "franka-panda", "policy": "vqbet", "state": "partial", "partial_scope": "Simulation only: Franka Kitchen (MuJoCo); no real-robot training or inference documented.", "framework": "vq-bet", "capabilities": { "inference_sim": { "state": "works", "notes": "VQ-BeT paper benchmarks on Franka Kitchen (state and image variants, goal- and non-goal-conditioned) with 566 human demonstrations. The paper's real-robot experiments used a Hello Robot Stretch, not a Panda.", "evidence": [ "ev-vqbet-paper-kitchen", "ev-vqbet-repo-kitchen" ] } }, "evidence": [ { "id": "ev-vqbet-paper-kitchen", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2403.03181v2", "date": "2024-06-28", "verified_on": "2026-10-25", "summary": "VQ-BeT paper (App. environments): Franka Kitchen is a simulated environment with a Franka Panda arm, used for VQ-BeT's simulation benchmarks.", "quote": "with a Franka Panda arm with a 7 dimensional action space and 566 human collected demonstrations." }, { "id": "ev-vqbet-repo-kitchen", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/jayLEE0301/vq_bet_official/blob/09d4851288ca5deaaa1ab367a208e520f8ee9a84/README.md", "date": "2024-02-28", "verified_on": "2026-10-25", "summary": "The official VQ-BeT repo ships training configs and pretrained weights for the simulated Kitchen environment.", "quote": "If you want to quickly see the performance of VQ-BeT on goal-cond Kitchen env without training it from scratch, please check the description below." } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/vqbet" }, { "robot": "franka-panda", "policy": "xvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "caveats": [ "The checkpoint's Panda identity comes from repo and robot_type naming (panda_ros_cartesian, panda_joint* states). The card is auto-generated with no hardware description." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-tinkybala-xvla" ] } }, "evidence": [ { "id": "ev-hub-tinkybala-xvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/TInkybala/Real_Panda_25epsdataset_xvla", "verified_on": "2026-10-24", "summary": "config.json type=xvla. train_config dataset TInkybala/Real_Panda_25eps_Screwdriver, whose meta/info.json robot_type=panda_ros_cartesian (68 episodes, one camera, Cartesian-velocity actions)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/franka-panda/xvla" }, { "robot": "hope-jr", "policy": "act", "state": "partial", "partial_scope": "Training ACT on data recorded with the Hope-JR hand (robot type hope_jr_hand) is documented upstream and labelled experimental; the arm is not covered. The page's evaluation step passes --policy.path to lerobot-record, but lerobot-record in v0.6.1 requires --teleop.type and has no policy option (policies are deployed with lerobot-rollout), so on-robot inference is not validly documented.", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "train": { "state": "works", "notes": "Documented for hope_jr_hand only, with a homunculus_glove teleoperator for recording; upstream calls Record/Replay/Train experimental.", "evidence": [ "ev-lerobot-hopejr-train", "ev-lerobot-hopejr-experimental" ] }, "teleop_record": { "state": "works", "evidence": [ "ev-lerobot-hopejr-experimental" ] } }, "caveats": [ "Upstream states that Record, Replay and Train with Hope-JR are still experimental." ], "evidence": [ { "id": "ev-lerobot-hopejr-train", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/hope_jr.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "The Hope-JR page trains an ACT policy on a hope_jr_hand dataset (outputs/train/hopejr_hand).", "quote": "--policy.type=act \\\n --output_dir=outputs/train/hopejr_hand" }, { "id": "ev-lerobot-hopejr-experimental", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/hope_jr.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "Upstream labels the Hope-JR record/replay/train workflow experimental; recording uses --robot.type=hope_jr_hand with a homunculus_glove teleoperator.", "quote": "Record, Replay and Train with Hope-JR is still experimental." }, { "id": "ev-lerobot-record-no-policy", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/src/lerobot/scripts/lerobot_record.py", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "In v0.6.1, RecordConfig has no policy field and raises an error when no teleoperator is given, pointing policy deployment to lerobot-rollout.", "quote": "For policy-based deployment, use lerobot-rollout instead." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/hope-jr/act" }, { "robot": "i2rt-yam", "policy": "act", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-hglll-act" ] } }, "evidence": [ { "id": "ev-hub-hglll-act", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/HGLLL/act-bimanual-yam-production-0320", "date": "2026-03-20", "verified_on": "2026-10-24", "summary": "config.json type=act; trained on HGLLL/bimanual-yam-0318 whose meta/info.json robot_type=bi_yam_follower (bimanual YAM)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/i2rt-yam/act" }, { "robot": "i2rt-yam", "policy": "diffusion", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-hglll-diffusion" ] } }, "evidence": [ { "id": "ev-hub-hglll-diffusion", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/HGLLL/diffusion-bimanual-yam-production-0321", "date": "2026-03-21", "verified_on": "2026-10-24", "summary": "config.json type=diffusion; trained on HGLLL/bimanual-yam-0318 whose meta/info.json robot_type=bi_yam_follower (bimanual YAM)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/i2rt-yam/diffusion" }, { "robot": "i2rt-yam", "policy": "groot-n1-7", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-helen-groot" ] } }, "evidence": [ { "id": "ev-hub-helen-groot", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/helen9975/groot-n1.7-molmoact-yam", "date": "2026-07-20", "verified_on": "2026-10-24", "summary": "config.json type=groot with base_model_path nvidia/GR00T-N1.7-3B; trained on allenai/MolmoAct2-BimanualYAM-Dataset whose meta/info.json robot_type=bi_yam_follower (bimanual YAM)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/i2rt-yam/groot-n1-7" }, { "robot": "i2rt-yam", "policy": "molmoact2", "state": "works", "framework": "molmoact2", "capabilities": { "train": { "state": "works", "notes": "Fine-tuned on the 720-hour MolmoAct2-BimanualYAM dataset and per-task on bimanual YAM data.", "evidence": [ "ev-molmoact2-paper-yam", "ev-molmoact2-bimanualyam-card" ] }, "inference_real": { "state": "works", "notes": "Bimanual YAM setup (two arms, top/left/right cameras). LeRobot v0.6.1 also documents a LeRobot-format checkpoint for lerobot-rollout, but LeRobot has no built-in YAM robot type, so that path needs a third-party YAM plugin.", "evidence": [ "ev-molmoact2-paper-yam-eval", "ev-molmoact2-yam-example", "ev-lerobot-molmoact2-doc" ] } }, "evidence": [ { "id": "ev-molmoact2-paper-yam", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2605.02881", "date": "2026-05-04", "verified_on": "2026-10-24", "summary": "MolmoAct2 paper: the MolmoAct2-BimanualYAM dataset was collected on the authors' bimanual YAM setup and used to train MolmoAct2.", "quote": "All data is collected on our custom bimanual YAM (Yet Another Manipulator) setup, shown in Figure 3." }, { "id": "ev-molmoact2-paper-yam-eval", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2605.02881", "date": "2026-05-04", "verified_on": "2026-10-24", "summary": "MolmoAct2 paper: real-world fine-tuning evaluation on bimanual and mobile bimanual YAM, 8 tasks x 50 trials (average success 50.1%).", "quote": "For real-world evaluation, we conduct a large-scale, systematic study on both a bimanual YAM setup and a mobile bimanual YAM setup, covering 8 tasks" }, { "id": "ev-molmoact2-bimanualyam-card", "tier": "upstream-documented", "type": "model-card", "url": "https://huggingface.co/allenai/MolmoAct2-BimanualYAM", "verified_on": "2026-10-24", "summary": "Official allenai/MolmoAct2-BimanualYAM checkpoint, fine-tuned for bimanual YAM inference (absolute joint-pose control).", "quote": "It is intended for both further fine-tuning and bimanual YAM policy inference." }, { "id": "ev-molmoact2-yam-example", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/allenai/molmoact2/blob/6070080a20321b4f498ab30f28e1d09ac465edb7/examples/yam/README.md", "verified_on": "2026-10-24", "summary": "Upstream molmoact2 repo ships a closed-loop bimanual YAM eval client that drives real YAM arms through the I2RT driver.", "quote": "Hardware-coupled example: it talks to real YAM arms over CAN (via `i2rt`) and" }, { "id": "ev-lerobot-molmoact2-doc", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/molmoact2.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 MolmoAct2 doc lists lerobot/MolmoAct2-BimanualYAM-LeRobot under Hardware Deployment (lerobot-rollout) with YAM camera keys.", "quote": "MolmoAct2-BimanualYAM-LeRobot | `top`, `left`, `right` | YAM 3-camera setup" } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/i2rt-yam/molmoact2" }, { "robot": "i2rt-yam", "policy": "openvla-oft", "state": "works-with-caveats", "framework": "openvla-oft", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested (evidence is a baseline in the MolmoAct2 paper, not the OpenVLA-OFT authors).", "The paper does not name the codebase used for this baseline; openvla-oft is recorded because it is the authors' reference implementation.", "Bimanual YAM setup; single-task fine-tunes averaged 35.5% success over 8 real-world tasks (50 trials each) in that study." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-molmoact2-paper-baselines", "ev-molmoact2-paper-c3" ] }, "inference_real": { "state": "works-with-caveats", "notes": "Average 35.5% success across 8 bimanual YAM tasks (Table 19 of the MolmoAct2 paper).", "evidence": [ "ev-molmoact2-paper-baselines", "ev-molmoact2-paper-c3" ] } }, "evidence": [ { "id": "ev-molmoact2-paper-baselines", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2605.02881", "date": "2026-05-04", "verified_on": "2026-10-24", "summary": "MolmoAct2 paper (Ai2): OpenVLA-OFT is one of four baselines fine-tuned per task and evaluated on a real bimanual YAM setup (8 tasks x 50 trials; OpenVLA-OFT average 35.5%).", "quote": "compare against four strong VLA and world-model baselines: Cosmos Policy (Kim et al., 2026a), X-VLA (Zheng et al., 2025), OpenVLA-OFT (Kim et al., 2025)" }, { "id": "ev-molmoact2-paper-c3", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2605.02881", "date": "2026-05-04", "verified_on": "2026-10-24", "summary": "Appendix C.3 (Real-world Bimanual YAM): every baseline was trained with the same protocol and its converged checkpoint evaluated on the robot.", "quote": "to ensure a fair comparison, we applied the identical training protocol to each of the four baseline policies. All policies were trained to convergence, and the resulting checkpoints were used for evaluation." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/i2rt-yam/openvla-oft" }, { "robot": "i2rt-yam", "policy": "pi05", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "Real-robot results come from a π0.5-DROID baseline in the MolmoAct2 paper (average 32.2% over 8 bimanual YAM tasks), which does not name the codebase used; the LeRobot checkpoints' cards do not report hardware rollouts.", "LeRobot has no built-in YAM robot type; the checkpoints use bi_yam_follower / yam_follower_bimanual types from third-party plugins." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-asethi-pi05", "ev-hub-nuffnuff-pi05" ] }, "inference_real": { "state": "works-with-caveats", "evidence": [ "ev-molmoact2-paper-c3" ] } }, "evidence": [ { "id": "ev-hub-asethi-pi05", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/ASethi04/pi05-BimanualYAM-oranges", "date": "2026-08-06", "verified_on": "2026-10-24", "summary": "config.json type=pi05 (fine-tuned from lerobot/pi05_base on a LeRobot fork whose pi05 code matches upstream main); trained on brandonyang/yam-vive-teleop whose meta/info.json robot_type=bi_yam_follower. Card: hardware rollouts pending." }, { "id": "ev-hub-nuffnuff-pi05", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/nuffnuff/pi05-yam-finetuned", "date": "2026-05-29", "verified_on": "2026-10-24", "summary": "config.json type=pi05; trained on nuffnuff/yam-master-v1 whose meta/info.json robot_type=yam_follower_bimanual." }, { "id": "ev-molmoact2-paper-c3", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2605.02881", "date": "2026-05-04", "verified_on": "2026-10-24", "summary": "MolmoAct2 paper appendix C.3: π0.5-DROID baseline fine-tuned per task and evaluated on real bimanual YAM, 8 tasks x 50 trials (Table 19 average 32.2%).", "quote": "to ensure a fair comparison, we applied the identical training protocol to each of the four baseline policies. All policies were trained to convergence, and the resulting checkpoints were used for evaluation." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/i2rt-yam/pi05" }, { "robot": "i2rt-yam", "policy": "smolvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-hcisbmm-smolvla" ] } }, "evidence": [ { "id": "ev-hub-hcisbmm-smolvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/hcisbmm/smolvla_yam_pink", "date": "2026-07-07", "verified_on": "2026-10-24", "summary": "config.json type=smolvla (base_model lerobot/smolvla_base); trained on hcisbmm/yam-pink-utensil-into-cup_20260707_180338 whose meta/info.json robot_type=bi_yam_follower (bimanual YAM)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/i2rt-yam/smolvla" }, { "robot": "i2rt-yam", "policy": "vla-jepa", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-helen-vlajepa", "ev-hub-jiafei-vlajepa" ] } }, "evidence": [ { "id": "ev-hub-helen-vlajepa", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/helen9975/vla-jepa-molmoact-yam-blocks-v2", "date": "2026-07-20", "verified_on": "2026-10-25", "summary": "config.json type=vla_jepa; train_config.json dataset allenai/MolmoAct2-BimanualYAM-Dataset (100000 steps, init lerobot/VLA-JEPA-Pretrain), whose meta/info.json robot_type=bi_yam_follower (bimanual YAM). Card reports only an open-loop MSE/MAE smoke test on 2 held-out episodes." }, { "id": "ev-hub-jiafei-vlajepa", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Jiafei1224/VLA-JEPA-MolmoAct2YAMData", "date": "2026-07-11", "verified_on": "2026-10-25", "summary": "config.json type=vla_jepa; train_config.json dataset allenai/MolmoAct2-BimanualYAM-Dataset (100000 steps, init lerobot/VLA-JEPA-Pretrain), whose meta/info.json robot_type=bi_yam_follower." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/i2rt-yam/vla-jepa" }, { "robot": "koch", "policy": "act", "state": "works", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "inference_real": { "state": "works", "notes": "The upstream quick-start rolls out the checkpoint lerobot/act_koch_real on koch_follower; that checkpoint was not publicly accessible on the Hub on 2026-10-24, so bring your own trained ACT checkpoint.", "evidence": [ "ev-lerobot-rollout-koch" ] }, "train": { "state": "works", "evidence": [ "ev-hub-ethancsl-act-koch", "ev-hub-sounderya-act-koch" ] } }, "evidence": [ { "id": "ev-lerobot-rollout-koch", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/inference.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot's policy-deployment page uses an ACT checkpoint on the Koch follower (koch_follower) as its lerobot-rollout quick-start example.", "quote": "--policy.path=lerobot/act_koch_real \\\n --robot.type=koch_follower" }, { "id": "ev-hub-ethancsl-act-koch", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/ethanCSL/act_koch_sorting_only_wrist", "verified_on": "2026-10-24", "summary": "config.json type=act; train_config.json dataset ethanCSL/svla_koch_sorting_n_stacking_wrist_camera whose meta/info.json robot_type=koch_follower (200 episodes)." }, { "id": "ev-hub-sounderya-act-koch", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Sounderya/act-koch-pickplace", "verified_on": "2026-10-24", "summary": "config.json type=act; train_config.json dataset Sounderya/koch_pickplace_v1 whose meta/info.json robot_type=koch_follower (210 episodes)." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/koch/act" }, { "robot": "koch", "policy": "diffusion", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-ssaito-diffusion-koch", "ev-hub-zliu157-diffusion-koch" ] } }, "evidence": [ { "id": "ev-hub-ssaito-diffusion-koch", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/ssaito/diffusion_koch_test", "verified_on": "2026-10-24", "summary": "config.json type=diffusion; train_config.json dataset ssaito/koch_test_104 whose meta/info.json robot_type=koch (50 episodes)." }, { "id": "ev-hub-zliu157-diffusion-koch", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/zliu157/diffusion_koch_i3r3", "verified_on": "2026-10-24", "summary": "config.json type=diffusion; train_config.json dataset zliu157/i3r3 whose meta/info.json robot_type=koch (60 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/koch/diffusion" }, { "robot": "koch", "policy": "openvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "openvla", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-phospho-openvla-koch" ] } }, "evidence": [ { "id": "ev-hub-phospho-openvla-koch", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/phospho-ai/openvla-koch", "verified_on": "2026-10-24", "summary": "Model card (base_model openvla/openvla-7b) states OpenVLA was fine-tuned on Koch robot datasets to control one; dataset_statistics.json has a single key koch_lerobot. Dataset robot_type not verifiable from the repo." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/koch/openvla" }, { "robot": "koch", "policy": "pi0-fast", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-ethancsl-pi0fast-koch" ] } }, "evidence": [ { "id": "ev-hub-ethancsl-pi0fast-koch", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/ethanCSL/pi0fast_koch_pick_n_place_vla_steering_height_test2", "verified_on": "2026-10-24", "summary": "config.json type=pi0_fast; train_config.json dataset ethanCSL/svla_koch_pick_n_place_vla_steering_height_test2 whose meta/info.json robot_type=koch_follower (61 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/koch/pi0-fast" }, { "robot": "koch", "policy": "pi0", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-rgarreta-pi0-koch", "ev-hub-ssaito-pi0-koch" ] } }, "evidence": [ { "id": "ev-hub-rgarreta-pi0-koch", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/rgarreta/pi0_koch_pick_place_lego_v8", "verified_on": "2026-10-24", "summary": "config.json type=pi0; train_config.json dataset rgarreta/koch_pick_place_lego_v8 whose meta/info.json robot_type=koch (50 episodes)." }, { "id": "ev-hub-ssaito-pi0-koch", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/ssaito/pi0_koch_test_6", "verified_on": "2026-10-24", "summary": "config.json type=pi0; train_config.json dataset ssaito/koch_test_104 whose meta/info.json robot_type=koch (50 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/koch/pi0" }, { "robot": "koch", "policy": "pi05", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-ethancsl-pi05-koch" ] } }, "evidence": [ { "id": "ev-hub-ethancsl-pi05-koch", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/ethanCSL/pi05_koch_sorting_only_wrist", "verified_on": "2026-10-24", "summary": "config.json type=pi05; train_config.json dataset ethanCSL/svla_koch_sorting_n_stacking_wrist_camera whose meta/info.json robot_type=koch_follower (200 episodes); fine-tuned from lerobot/pi05_base." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/koch/pi05" }, { "robot": "koch", "policy": "smolvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-ethancsl-smolvla-koch", "ev-hub-sounderya-smolvla-koch" ] } }, "evidence": [ { "id": "ev-hub-ethancsl-smolvla-koch", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/ethanCSL/svla_koch_sorting_only_wrist", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config.json dataset ethanCSL/svla_koch_sorting_n_stacking_wrist_camera whose meta/info.json robot_type=koch_follower (200 episodes); fine-tuned from lerobot/smolvla_base." }, { "id": "ev-hub-sounderya-smolvla-koch", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Sounderya/smolvla-koch-pickplace", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config.json dataset Sounderya/koch_pickplace_v1 whose meta/info.json robot_type=koch_follower (210 episodes); fine-tuned from lerobot/smolvla_base." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/koch/smolvla" }, { "robot": "lekiwi", "policy": "act", "state": "works", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "inference_real": { "state": "works", "notes": "Upstream example examples/lekiwi/evaluate.py loads an ACTPolicy and runs it against LeKiwiClient; the script itself notes that lerobot-rollout is the production deployment path.", "evidence": [ "ev-lerobot-lekiwi-doc", "ev-lerobot-lekiwi-evaluate" ] }, "train": { "state": "works", "evidence": [ "ev-hub-pepijn-act-lekiwi", "ev-hub-masatoka-act-lekiwi" ] } }, "evidence": [ { "id": "ev-lerobot-lekiwi-doc", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/lekiwi.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "The LeKiwi page points policy evaluation at the examples/lekiwi/evaluate.py script.", "quote": "To evaluate your policy run the `evaluate.py` API example, make sure to change `remote_ip`, `port`, model.." }, { "id": "ev-lerobot-lekiwi-evaluate", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/examples/lekiwi/evaluate.py", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "The LeKiwi evaluation example instantiates LeKiwiClient and an ACTPolicy loaded from the Hub, then runs the policy on the robot.", "quote": "robot = LeKiwiClient(robot_config)\n\n # Create policy\n policy = ACTPolicy.from_pretrained(HF_MODEL_ID)" }, { "id": "ev-hub-pepijn-act-lekiwi", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/pepijn223/act_lekiwi", "verified_on": "2026-10-24", "summary": "config.json type=act; train_config.json dataset pepijn223/lekiwi_pen whose meta/info.json robot_type=lekiwi (40 episodes)." }, { "id": "ev-hub-masatoka-act-lekiwi", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/masato-ka/act_lekiwi_dounut_serve_full", "verified_on": "2026-10-24", "summary": "config.json type=act; train_config.json dataset masato-ka/lekiwi_dounut_serve_full whose meta/info.json robot_type=lekiwi_client (60 episodes)." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/lekiwi/act" }, { "robot": "lekiwi", "policy": "diffusion", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-liosgr-diffusion", "ev-hub-wiscohumanoids-diffusion" ] } }, "evidence": [ { "id": "ev-hub-liosgr-diffusion", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/lioSgr/dp_pick_toy_elephant", "date": "2026-09-18", "verified_on": "2026-10-25", "summary": "config.json type=diffusion; train_config.json dataset lioSgr/pick_toy_elephant (100000 steps), whose meta/info.json robot_type=lekiwi_client (50 episodes; arm joints plus x/y/theta base velocities; front + wrist cameras)." }, { "id": "ev-hub-wiscohumanoids-diffusion", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/wiscohumanoids/pick_up_single_colored_cube_into_tape_v7_diffusion_h128", "date": "2026-04-15", "verified_on": "2026-10-25", "summary": "config.json type=diffusion; train_config.json dataset wiscohumanoids/pick_up_single_colored_cube_into_tape_v7 (80000 steps), whose meta/info.json robot_type=lekiwi_client (226 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/lekiwi/diffusion" }, { "robot": "lekiwi", "policy": "pi05", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-tinjyuu-pi05-lekiwi", "ev-hub-alexppppp-pi05-lekiwi" ] } }, "evidence": [ { "id": "ev-hub-tinjyuu-pi05-lekiwi", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/tinjyuu/pi05_lekiwi_exp1", "verified_on": "2026-10-24", "summary": "config.json type=pi05; train_config.json dataset tinjyuu/lekiwi-pink-plush-box1 whose meta/info.json robot_type=lekiwi_client (30 episodes)." }, { "id": "ev-hub-alexppppp-pi05-lekiwi", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/alexppppp/pi05_base_lekiwi_new_gripper_multitask-3ksteps", "verified_on": "2026-10-24", "summary": "config.json type=pi05; train_config.json dataset alexppppp/lekiwi_new_gripper_multitask whose meta/info.json robot_type=lekiwi_client (29 episodes); initialised from a lerobot/pi05_base snapshot." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/lekiwi/pi05" }, { "robot": "lekiwi", "policy": "smolvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-norips-smolvla-lekiwi", "ev-hub-fruityapplebottom-smolvla-lekiwi" ] } }, "evidence": [ { "id": "ev-hub-norips-smolvla-lekiwi", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/norips/smolvla_lekiwi_socks", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config.json dataset norips/lekiwi_socks whose meta/info.json robot_type=lekiwi_client (60 episodes)." }, { "id": "ev-hub-fruityapplebottom-smolvla-lekiwi", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/fruityapplebottom/smolvla_toy_in_cup_lekiwi_mobile", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config.json dataset fruityapplebottom/toyincup50_b whose meta/info.json robot_type=lekiwi_client (80 episodes); fine-tuned from lerobot/smolvla_base." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/lekiwi/smolvla" }, { "robot": "mobile-aloha", "policy": "act", "state": "works", "framework": "act-plus-plus", "caveats": [ "Documented with the Mobile ALOHA authors' codebase (MarkFzp/act-plus-plus, co-training with static ALOHA data), not LeRobot; LeRobot's converted lerobot/aloha_mobile_* datasets carry 14-dim arm actions only (no base velocity)." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-mobile-aloha-paper", "ev-act-plus-plus" ] }, "inference_real": { "state": "works", "evidence": [ "ev-mobile-aloha-paper" ] } }, "evidence": [ { "id": "ev-mobile-aloha-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2401.02117", "date": "2024-01-04", "verified_on": "2026-10-24", "summary": "Mobile ALOHA paper trains ACT (co-trained with static ALOHA data) for whole-body mobile manipulation tasks on Mobile ALOHA, e.g. 95% success on Wipe Wine with co-training.", "quote": "We train two recent imitation learning methods, Diffusion Policy [18] and VINN [63], with Mobile ALOHA in addition to ACT." }, { "id": "ev-act-plus-plus", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/MarkFzp/act-plus-plus/blob/26bab0789d05b7496bacef04f5c6b2541a4403b5/README.md", "date": "2024-01-11", "verified_on": "2026-10-24", "summary": "Authors' repo implementing ACT, Diffusion Policy and VINN with co-training for Mobile ALOHA; real-robot use requires the MarkFzp/mobile-aloha stack.", "quote": "This repo contains the implementation of ACT, Diffusion Policy and VINN, together with 2 simulated environments:" } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/mobile-aloha/act" }, { "robot": "mobile-aloha", "policy": "diffusion", "state": "works-with-caveats", "framework": "act-plus-plus", "caveats": [ "With 50 demonstrations Diffusion Policy was less precise than ACT on Wipe Wine (65% success); the authors attribute this to too few demonstrations.", "Documented with the Mobile ALOHA authors' codebase (MarkFzp/act-plus-plus, DDIM scheduler, co-training with static ALOHA data), not LeRobot's diffusion policy." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-mobile-aloha-paper", "ev-act-plus-plus" ] }, "inference_real": { "state": "works", "evidence": [ "ev-mobile-aloha-paper" ] } }, "evidence": [ { "id": "ev-mobile-aloha-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2401.02117", "date": "2024-01-04", "verified_on": "2026-10-24", "summary": "Mobile ALOHA paper (Sec. 6.2) trains Diffusion Policy on Mobile ALOHA: 100% on Push Chairs with co-training, 65% on Wipe Wine.", "quote": "Overall, Diffusion Policy performs similarly to ACT on Push Chairs, both obtaining 100% with co-training. For Wipe Wine, we observe worse performance with diffusion at 65% success." }, { "id": "ev-act-plus-plus", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/MarkFzp/act-plus-plus/blob/26bab0789d05b7496bacef04f5c6b2541a4403b5/README.md", "date": "2024-01-11", "verified_on": "2026-10-24", "summary": "Authors' repo implementing ACT, Diffusion Policy and VINN with co-training for Mobile ALOHA.", "quote": "This repo contains the implementation of ACT, Diffusion Policy and VINN, together with 2 simulated environments:" } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/mobile-aloha/diffusion" }, { "robot": "mobile-aloha", "policy": "pi0", "state": "partial", "partial_scope": "Only a misconfigured zero-shot pi0_base run is documented: it executed on Mobile ALOHA hardware but produced no task behavior. No fine-tuned or correctly configured Mobile ALOHA run is publicly documented.", "framework": "openpi", "capabilities": { "inference_real": { "state": "partial", "notes": "Issue #332: pi0_base served through openpi's static ALOHA runtime (--env ALOHA) with the static 'trossen' norm stats; the robot only moved slightly in place. A Physical Intelligence contributor said the 'trossen_mobile' stats should be used, zero-shot success is not expected, and fine-tuning is recommended.", "evidence": [ "ev-openpi-issue-332", "ev-openpi-issue-332-reply", "ev-openpi-norm-stats-mobile-aloha" ] } }, "evidence": [ { "id": "ev-openpi-issue-332", "tier": "community-reported", "type": "github-issue", "url": "https://github.com/Physical-Intelligence/openpi/issues/332", "date": "2025-02-26", "verified_on": "2026-10-25", "summary": "User ran pi0_base zero-shot on Mobile ALOHA via examples/aloha_real (--env ALOHA, 'pick a cup'); checkpoint and 'trossen' norm stats loaded, but the robot only moved slightly and never approached the cup.", "quote": "The robot moves slightly in place but does not go towards the cup." }, { "id": "ev-openpi-issue-332-reply", "tier": "community-reported", "type": "github-issue", "url": "https://github.com/Physical-Intelligence/openpi/issues/332#issuecomment-2702822480", "date": "2025-03-06", "verified_on": "2026-10-25", "summary": "Physical Intelligence contributor: zero-shot on a new task is not expected to work, Mobile ALOHA is untested outside their office, fine-tune on task data, and load 'trossen_mobile' rather than 'trossen' norm stats.", "quote": "I do not expect pi0 to work zero-shot on a new task and we have not tested performance on mobile aloha's outside of our office." }, { "id": "ev-openpi-norm-stats-mobile-aloha", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/Physical-Intelligence/openpi/blob/070759fae105d05e958efe78fe6fb08631f23da5/docs/norm_stats.md", "date": "2025-06-20", "verified_on": "2026-10-25", "summary": "Supporting only (not deployment evidence): openpi publishes pi0_base pre-training normalization statistics for Mobile ALOHA (asset 'trossen_mobile').", "quote": "Mobile ALOHA | Mobile version of ALOHA mounted on a Slate base | trossen_mobile" } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/mobile-aloha/pi0" }, { "robot": "mobile-aloha", "policy": "smolvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-naonaon-smolvla-cabinet", "ev-hub-qualia-smolvla-elevator" ] } }, "evidence": [ { "id": "ev-hub-naonaon-smolvla-cabinet", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/naonaon/smolvla_aloha_cabinet", "date": "2025-12-27", "verified_on": "2026-10-25", "summary": "config.json type=smolvla (init lerobot/smolvla_base); train_config.json dataset lerobot/aloha_mobile_cabinet (50000 steps), whose meta/info.json robot_type=aloha (85 episodes at 50 fps; one of LeRobot's Mobile ALOHA datasets)." }, { "id": "ev-hub-qualia-smolvla-elevator", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/qualiaadmin/6ed43167-aed0-4d17-ac0d-801cb4bab10e", "date": "2026-01-15", "verified_on": "2026-10-25", "summary": "config.json type=smolvla; train_config.json dataset lerobot/aloha_mobile_elevator, whose meta/info.json robot_type=aloha (20 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/mobile-aloha/smolvla" }, { "robot": "moss-v1", "policy": "act", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "LeRobot's `moss` robot type and its tutorial were removed in the hardware API redesign (PR #777, June 2025); Moss v1 is not supported by LeRobot v0.6.1, so the documented workflow needs a pre-redesign commit." ], "capabilities": { "teleop_record": { "state": "works", "evidence": [ "ev-lerobot-moss-doc" ] }, "train": { "state": "works", "evidence": [ "ev-lerobot-moss-doc" ] }, "inference_real": { "state": "works", "evidence": [ "ev-lerobot-moss-doc" ] } }, "evidence": [ { "id": "ev-lerobot-moss-doc", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/bed90e3a41c43758c619dba66158ddd5798d361a/examples/11_use_moss.md", "verified_on": "2026-10-24", "summary": "Official LeRobot Moss v1 tutorial (last version before removal, commit bed90e3, May 2025): calibrate/teleop/record with --robot.type=moss, train with --policy.type=act, evaluate the ACT checkpoint on the robot via --control.policy.path.", "quote": "This tutorial explains how to use [Moss v1](https://github.com/jess-moss/moss-robot-arms) with LeRobot." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/moss-v1/act" }, { "robot": "omx", "policy": "act", "state": "works", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "train": { "state": "works", "evidence": [ "ev-lerobot-omx-example", "ev-hub-maximellerbach-omx-act", "ev-hub-imsubin-omx-act" ] }, "inference_real": { "state": "works", "notes": "Rolled out with lerobot-rollout --robot.type=omx_follower (base or sentry strategy) using two OpenCV cameras (wrist + top-down). Upstream describes the example as WIP and specific to its own setup.", "evidence": [ "ev-lerobot-omx-example" ] }, "teleop_record": { "state": "works", "evidence": [ "ev-lerobot-omx-example" ] } }, "evidence": [ { "id": "ev-lerobot-omx-example", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/examples/omx/README.md", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot's OMX Follower example records a cube pick-and-place dataset, trains ACT with lerobot-train and deploys it with lerobot-rollout on omx_follower.", "quote": "To train a simple `ACT` policy on the collected dataset, you can use the `lerobot-train` CLI:" }, { "id": "ev-hub-maximellerbach-omx-act", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/maximellerbach/omx_pickandplace_act", "verified_on": "2026-10-24", "summary": "Checkpoint linked from the upstream OMX example. config.json type=act; train_config.json dataset maximellerbach/omx_autocollect whose meta/info.json robot_type=omx_follower (50 episodes)." }, { "id": "ev-hub-imsubin-omx-act", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/IMsubin/pick_doll_put_pot_omx_act2", "verified_on": "2026-10-24", "summary": "config.json type=act; train_config.json dataset IMsubin/pick_doll_put_pot_omx whose meta/info.json robot_type=omx_follower (100 episodes)." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/act" }, { "robot": "omx", "policy": "diffusion", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-mingminng-diffusion-omx", "ev-hub-mingminng-diffusion-omx-2" ] } }, "evidence": [ { "id": "ev-hub-mingminng-diffusion-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Mingminng/diffusion_omx_pick_color_motor_3_V1", "verified_on": "2026-10-24", "summary": "config.json type=diffusion; train_config.json dataset Mingminng/omx_pick_color_motor_3 whose meta/info.json robot_type=omx_follower (60 episodes)." }, { "id": "ev-hub-mingminng-diffusion-omx-2", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Mingminng/diffusion_omx_pick_motor10", "verified_on": "2026-10-24", "summary": "config.json type=diffusion; train_config.json dataset Mingminng/omx_pick_motor7 whose meta/info.json robot_type=omx_follower (10 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/diffusion" }, { "robot": "omx", "policy": "eo1", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-pepijn-eo1" ] } }, "evidence": [ { "id": "ev-hub-pepijn-eo1", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/pepijn223/eo1_super_poulain_draft", "date": "2026-05-01", "verified_on": "2026-10-25", "summary": "config.json type=eo1; train_config.json dataset imstevenpmwork/super_poulain_draft (10000 steps), whose meta/info.json robot_type=omx_follower (50 episodes). Repo is labelled a draft; auto-generated card, no results." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/eo1" }, { "robot": "omx", "policy": "evo1", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Physical-run evidence is post-0.6.1, unreleased LeRobot main (remote-inference docs, PR #4937); no released LeRobot version documents it.", "No source names the robot for the EVO1 run: upstream lists the checkpoint imstevenpmwork/super_chatton_evo1_stage1_pro under Remote HW and says EVO1 checkpoints ran on real robots; OMX is attributed only because that checkpoint was trained solely on an omx_follower dataset.", "No task success rate, trial count or latency is published for this run." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-maximellerbach-evo1-omx", "ev-hub-imstevenpmwork-evo1-omx" ] }, "inference_real": { "state": "works-with-caveats", "notes": "Remote-inference hardware run of an OMX-trained EVO1 checkpoint per LeRobot main docs; the robot is not named by the source.", "evidence": [ "ev-lerobot-main-remote-readme", "ev-lerobot-main-remote-guide", "ev-hub-imstevenpmwork-evo1-omx" ] } }, "evidence": [ { "id": "ev-hub-maximellerbach-evo1-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/maximellerbach/omx_pickandplace_evo1", "verified_on": "2026-10-24", "summary": "config.json type=evo1; train_config.json dataset maximellerbach/omx_pickandplace whose meta/info.json robot_type=omx_follower (50 episodes); 5000 steps." }, { "id": "ev-hub-imstevenpmwork-evo1-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/imstevenpmwork/super_chatton_evo1_stage1_pro/blob/80cd23cdd4f20ee1d4203888d5d9fa088704d8aa/README.md", "date": "2026-07-02", "verified_on": "2026-10-25", "versions": { "lerobot": "0.5.2" }, "summary": "config.json type=evo1 (5000 steps); train_config.json dataset imstevenpmwork/super_chatton whose meta/info.json robot_type=omx_follower (100 episodes, front + wrist cameras, 6-D action). Auto-generated card; its rollout command is a placeholder template." }, { "id": "ev-lerobot-main-remote-readme", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/d3793b7e36522321c1c455f0dc4826db3793dceb/src/lerobot/remote_inference/README.md", "date": "2026-10-14", "verified_on": "2026-10-25", "versions": { "lerobot": "main@d3793b7e36522321c1c455f0dc4826db3793dceb" }, "summary": "Post-0.6.1 (unreleased main) remote-inference reference: the policy-support table's 'Execution evidence' column lists imstevenpmwork/super_chatton_evo1_stage1_pro under Remote HW. It names the checkpoint, not the robot.", "quote": "Remote HW: `imstevenpmwork/super_chatton_evo1_stage1_pro`." }, { "id": "ev-lerobot-main-remote-guide", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/d3793b7e36522321c1c455f0dc4826db3793dceb/docs/source/remote_inference.mdx", "date": "2026-10-14", "verified_on": "2026-10-25", "versions": { "lerobot": "main@d3793b7e36522321c1c455f0dc4826db3793dceb" }, "summary": "Post-0.6.1 (unreleased main) remote-inference user guide: states EVO1 checkpoints have run on real robots through the remote-inference path (robots not named).", "quote": "ACT, SmolVLA, XVLA, EVO1 and LaWAM checkpoints have run on real robots through this path." } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/evo1" }, { "robot": "omx", "policy": "fastwam", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-maximellerbach-fastwam-omx" ] } }, "evidence": [ { "id": "ev-hub-maximellerbach-fastwam-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/maximellerbach/omx_multicubes_fastwam", "verified_on": "2026-10-24", "summary": "config.json type=fastwam; train_config.json dataset maximellerbach/omx_multicubes whose meta/info.json robot_type=omx_follower (176 episodes); pretrained from lerobot/fastwam_base." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/fastwam" }, { "robot": "omx", "policy": "groot-n1-7", "state": "works-with-caveats", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "Trained and served with NVIDIA's Isaac-GR00T stack (NEW_EMBODIMENT tag, GR00T policy server), not LeRobot's groot policy; the dataset uses ROBOTIS's robot_type string omx_f rather than LeRobot's omx_follower.", "The authors report that the single top-down-camera model closed the gripper before reaching the object during real-robot deployment (height perception); their 2-camera successor adds a wrist camera to address it." ], "framework": "isaac-groot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-pseudolab-groot-omx-2cam", "ev-hub-pseudolab-groot-omx" ] }, "inference_real": { "state": "works-with-caveats", "notes": "Only qualitative real-robot feedback is given (no success rates); reported metrics are open-loop MSE.", "evidence": [ "ev-hub-pseudolab-groot-omx-2cam" ] } }, "evidence": [ { "id": "ev-hub-pseudolab-groot-omx-2cam", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/pseudolab/GR00T-N1.7-3B-OMX-PickupDolls-2Cam", "verified_on": "2026-10-24", "summary": "Model card: fine-tune of nvidia/GR00T-N1.7-3B (NEW_EMBODIMENT, frozen backbone) for the 'ROBOTIS Open Manipulator-X' (card wording) 'Pick up dolls' task on pseudolab/omx_f_PickUpDollWith2Cam, whose meta/info.json robot_type=omx_f (94 episodes); state is 5 arm joints + gripper. The card cites real-robot deployment of the 1-camera predecessor." }, { "id": "ev-hub-pseudolab-groot-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/pseudolab/GR00T-N1.7-3B-OMX-PickupDolls", "verified_on": "2026-10-24", "summary": "Model card: GR00T N1.7-3B fine-tuned (NEW_EMBODIMENT) for the Open Manipulator-X 5-DoF + gripper doll pick-up task on 75 teleoperated episodes plus Cosmos-Transfer2.5 augmentation; open-loop MSE only." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/groot-n1-7" }, { "robot": "omx", "policy": "lingbot-va", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-maximellerbach-lingbotva-omx" ] } }, "evidence": [ { "id": "ev-hub-maximellerbach-lingbotva-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/maximellerbach/omx_multicubes_lingbot_va", "verified_on": "2026-10-24", "summary": "config.json type=lingbot_va; train_config.json dataset maximellerbach/omx_multicubes whose meta/info.json robot_type=omx_follower (176 episodes); pretrained from lerobot/lingbot_va_base." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/lingbot-va" }, { "robot": "omx", "policy": "molmoact2", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-maximellerbach-molmoact2-omx" ] } }, "evidence": [ { "id": "ev-hub-maximellerbach-molmoact2-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/maximellerbach/omx_pickandplace_molmoact2", "verified_on": "2026-10-24", "summary": "config.json type=molmoact2; train_config.json dataset maximellerbach/omx_pickandplace whose meta/info.json robot_type=omx_follower (50 episodes); 5000 steps." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/molmoact2" }, { "robot": "omx", "policy": "multitask-dit", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "caveats": [ "LeRobot main's (post-0.6.1) remote-inference runner does not yet support Multi-task DiT: its history/queue preparation needs adaptation to the current-observation contract." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-imstevenpmwork-dit-omx" ] } }, "evidence": [ { "id": "ev-hub-imstevenpmwork-dit-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/imstevenpmwork/dit_super_chatton_4549-ema/blob/9360c6071fb6dc944dc4db0612268d366cf1ac67/README.md", "date": "2026-09-08", "verified_on": "2026-10-25", "versions": { "lerobot": "0.6.2 (unreleased dev version per card; latest release is 0.6.1)" }, "summary": "config.json type=multi_task_dit (20000 steps); train_config.json dataset imstevenpmwork/super_chatton whose meta/info.json robot_type=omx_follower (100 episodes, front + wrist cameras). Auto-generated card with a placeholder rollout command and no evaluation results.", "quote": "- **Robot type:** `omx_follower`" }, { "id": "ev-lerobot-main-remote-readme", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/d3793b7e36522321c1c455f0dc4826db3793dceb/src/lerobot/remote_inference/README.md", "date": "2026-10-14", "verified_on": "2026-10-25", "versions": { "lerobot": "main@d3793b7e36522321c1c455f0dc4826db3793dceb" }, "summary": "Post-0.6.1 (unreleased main) remote-inference reference: Diffusion, Multi-task DiT and VQ-BeT need sampling/preparation and horizon adaptation before the remote runner can serve them.", "quote": "Default history and/or per-step queue preparation do not satisfy the direct current-observation contract." } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/multitask-dit" }, { "robot": "omx", "policy": "pi0", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-yaswanth-pi0-omx" ] } }, "evidence": [ { "id": "ev-hub-yaswanth-pi0-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/yaswanth8390/omx_pi0", "verified_on": "2026-10-24", "summary": "config.json type=pi0; train_config.json dataset yaswanth8390/merged whose meta/info.json robot_type=omx_follower (100 episodes); pretrained from lerobot/pi0_base." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/pi0" }, { "robot": "omx", "policy": "pi05", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-linhanwang-pi05-omx", "ev-hub-masatoka-pi05-omx" ] } }, "evidence": [ { "id": "ev-hub-linhanwang-pi05-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/LinhanWang/pi05_omx_three_tasks", "verified_on": "2026-10-24", "summary": "Checkpoint folder 010000: config.json type=pi05; train_config.json dataset LinhanWang/omx_three_tasks whose meta/info.json robot_type=omx_follower (150 episodes); pretrained from lerobot/pi05_base." }, { "id": "ev-hub-masatoka-pi05-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/masato-ka/pi05-omx-transfer-cube-v1-0809", "verified_on": "2026-10-24", "summary": "config.json type=pi05; train_config.json dataset masato-ka/omx-transfer-cube-v1_20260809_231300 whose meta/info.json robot_type=omx_follower (25 episodes); pretrained from lerobot/pi05_base." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/pi05" }, { "robot": "omx", "policy": "smolvla", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Physical-run evidence is post-0.6.1, unreleased LeRobot main (remote-inference docs, PR #4937); no released LeRobot version documents it.", "The upstream support table names the checkpoint (imstevenpmwork/super_chatton_smolvla), not the robot; OMX comes from that checkpoint's training dataset (robot_type=omx_follower) and from the PR's 'Local OMX/SmolVLA' hardware note.", "No task success rate, trial count or latency is published for this run." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-adarshkorde-smolvla-omx", "ev-hub-imsubin-smolvla-omx", "ev-hub-imstevenpmwork-smolvla-omx" ] }, "inference_real": { "state": "works-with-caveats", "notes": "Remote inference and local guided RTC on hardware per LeRobot main docs and PR #4937; no success metrics.", "evidence": [ "ev-lerobot-main-remote-readme", "ev-lerobot-main-remote-guide", "ev-lerobot-pr-4937" ] } }, "evidence": [ { "id": "ev-hub-adarshkorde-smolvla-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/AdarshKorde/omx_smolvla_green_cube_20000", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config.json dataset AdarshKorde/omx_pick_green_cube whose meta/info.json robot_type=omx_follower (50 episodes); fine-tuned from lerobot/smolvla_base." }, { "id": "ev-hub-imsubin-smolvla-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/IMsubin/pick_doll_put_pot_omx_smolvla5", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config.json dataset IMsubin/pick_doll_put_pot_omx whose meta/info.json robot_type=omx_follower (100 episodes)." }, { "id": "ev-hub-imstevenpmwork-smolvla-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/imstevenpmwork/super_chatton_smolvla/blob/1dfe94c5c61f330cb513c6f752095c2de3009ebb/README.md", "date": "2026-06-16", "verified_on": "2026-10-25", "versions": { "lerobot": "0.5.2" }, "summary": "config.json type=smolvla (fine-tuned from lerobot/smolvla_base, 20000 steps); train_config.json dataset imstevenpmwork/super_chatton whose meta/info.json robot_type=omx_follower (100 episodes, front + wrist cameras). Auto-generated card; its rollout command is a placeholder template." }, { "id": "ev-lerobot-main-remote-readme", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/d3793b7e36522321c1c455f0dc4826db3793dceb/src/lerobot/remote_inference/README.md", "date": "2026-10-14", "verified_on": "2026-10-25", "versions": { "lerobot": "main@d3793b7e36522321c1c455f0dc4826db3793dceb" }, "summary": "Post-0.6.1 (unreleased main) remote-inference reference: the policy-support table's 'Execution evidence' column lists imstevenpmwork/super_chatton_smolvla under Remote HW and local guided RTC. It names the checkpoint, not the robot.", "quote": "Remote HW and local guided RTC: `imstevenpmwork/super_chatton_smolvla`." }, { "id": "ev-lerobot-main-remote-guide", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/d3793b7e36522321c1c455f0dc4826db3793dceb/docs/source/remote_inference.mdx", "date": "2026-10-14", "verified_on": "2026-10-25", "versions": { "lerobot": "main@d3793b7e36522321c1c455f0dc4826db3793dceb" }, "summary": "Post-0.6.1 (unreleased main) remote-inference user guide: states SmolVLA checkpoints have run on real robots through the remote-inference path (robots not named).", "quote": "ACT, SmolVLA, XVLA, EVO1 and LaWAM checkpoints have run on real robots through this path." }, { "id": "ev-lerobot-pr-4937", "tier": "community-reported", "type": "github-pr", "url": "https://github.com/huggingface/lerobot/pull/4937", "date": "2026-10-13", "verified_on": "2026-10-25", "versions": { "lerobot": "main@ae2b57f12223c8c5184fcd814eb63cd9ce7de203" }, "summary": "Remote-inference PR by imstevenpmwork (LeRobot collaborator; merged into main as ae2b57f, after v0.6.1) reports hardware rollouts with SmolVLA and names OMX for the local SmolVLA guided-RTC checks.", "quote": "Local OMX/SmolVLA guided RTC passed recovery and expiry checks." } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/smolvla" }, { "robot": "omx", "policy": "vla-jepa", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-maximellerbach-vlajepa-omx", "ev-hub-dawity-vlajepa-omx" ] } }, "evidence": [ { "id": "ev-hub-maximellerbach-vlajepa-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/maximellerbach/omx_pickandplace_vla_jepa", "verified_on": "2026-10-24", "summary": "config.json type=vla_jepa; train_config.json dataset maximellerbach/omx_pickandplace whose meta/info.json robot_type=omx_follower (50 episodes); pretrained from maximellerbach/VLA-JEPA-Pretrain." }, { "id": "ev-hub-dawity-vlajepa-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/dawity/vla_jepa_omx", "verified_on": "2026-10-24", "summary": "config.json type=vla_jepa; train_config.json dataset dawity/vla_fresh_data_drawer3 whose meta/info.json robot_type=omx_follower (150 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/vla-jepa" }, { "robot": "omx", "policy": "wall-oss", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-imstevenpmwork-wallx" ] } }, "evidence": [ { "id": "ev-hub-imstevenpmwork-wallx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/imstevenpmwork/wall_x_super_chatton", "date": "2026-07-17", "verified_on": "2026-10-25", "summary": "config.json type=wall_x; train_config.json dataset imstevenpmwork/super_chatton (8000 steps), whose meta/info.json robot_type=omx_follower (100 episodes). Card has no rollout results." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/wall-oss" }, { "robot": "omx", "policy": "xvla", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Physical-run evidence is post-0.6.1, unreleased LeRobot main (remote-inference docs, PR #4937); no released LeRobot version documents it.", "The upstream support table names the checkpoint (imstevenpmwork/xvla_super_chatton_2), not the robot; OMX comes from that checkpoint's training dataset (robot_type=omx_follower) and from the PR's 'OMX/XVLA checks' hardware note.", "Upstream notes X-VLA runs chunk-only through the remote-inference runner; no task success rate, trial count or latency is published." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-mingminng-xvla-omx", "ev-hub-mingminng-xvla-omx-2", "ev-hub-imstevenpmwork-xvla-omx" ] }, "inference_real": { "state": "works-with-caveats", "notes": "Remote-inference hardware rollouts (pause/resume, starvation recovery, DAgger-style recording) per LeRobot main docs and PR #4937; no success metrics.", "evidence": [ "ev-lerobot-main-remote-readme", "ev-lerobot-main-remote-guide", "ev-lerobot-pr-4937" ] } }, "evidence": [ { "id": "ev-hub-mingminng-xvla-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Mingminng/xvla_omx_pick_color_motor_3_V1", "verified_on": "2026-10-24", "summary": "config.json type=xvla; train_config.json dataset Mingminng/omx_pick_color_motor_3 whose meta/info.json robot_type=omx_follower (60 episodes); fine-tuned from lerobot/xvla-base." }, { "id": "ev-hub-mingminng-xvla-omx-2", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Mingminng/xvla_omx_pick_motor2", "verified_on": "2026-10-24", "summary": "config.json type=xvla; train_config.json dataset Mingminng/omx_pick_motor7 whose meta/info.json robot_type=omx_follower (10 episodes); fine-tuned from lerobot/xvla-base." }, { "id": "ev-hub-imstevenpmwork-xvla-omx", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/imstevenpmwork/xvla_super_chatton_2/blob/36fbf138211f3084e8bcd1c1c9a506ab522bb85d/README.md", "date": "2026-07-20", "verified_on": "2026-10-25", "versions": { "lerobot": "0.6.1" }, "summary": "config.json type=xvla (fine-tuned from lerobot/xvla-base, 10000 steps); train_config.json dataset imstevenpmwork/super_chatton whose meta/info.json robot_type=omx_follower (100 episodes, front + wrist cameras). Auto-generated card; its rollout command is a placeholder template." }, { "id": "ev-lerobot-main-remote-readme", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/d3793b7e36522321c1c455f0dc4826db3793dceb/src/lerobot/remote_inference/README.md", "date": "2026-10-14", "verified_on": "2026-10-25", "versions": { "lerobot": "main@d3793b7e36522321c1c455f0dc4826db3793dceb" }, "summary": "Post-0.6.1 (unreleased main) remote-inference reference: the policy-support table's 'Execution evidence' column lists imstevenpmwork/xvla_super_chatton_2 under Remote HW. It names the checkpoint, not the robot.", "quote": "Remote HW: `imstevenpmwork/xvla_super_chatton_2`." }, { "id": "ev-lerobot-main-remote-guide", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/d3793b7e36522321c1c455f0dc4826db3793dceb/docs/source/remote_inference.mdx", "date": "2026-10-14", "verified_on": "2026-10-25", "versions": { "lerobot": "main@d3793b7e36522321c1c455f0dc4826db3793dceb" }, "summary": "Post-0.6.1 (unreleased main) remote-inference user guide: states XVLA checkpoints have run on real robots through the remote-inference path (robots not named).", "quote": "ACT, SmolVLA, XVLA, EVO1 and LaWAM checkpoints have run on real robots through this path." }, { "id": "ev-lerobot-pr-4937", "tier": "community-reported", "type": "github-pr", "url": "https://github.com/huggingface/lerobot/pull/4937", "date": "2026-10-13", "verified_on": "2026-10-25", "versions": { "lerobot": "main@ae2b57f12223c8c5184fcd814eb63cd9ce7de203" }, "summary": "Remote-inference PR by imstevenpmwork (LeRobot collaborator; merged into main as ae2b57f, after v0.6.1) names OMX/XVLA for its hardware checks of pause/resume, starvation recovery and recording strategies.", "quote": "OMX/XVLA checks covered pause/resume, starvation recovery/expiry and Sentry, Highlight, Episodic and corrections-only DAgger recording; uploaded datasets and videos were inspected." } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/omx/xvla" }, { "robot": "openarm", "policy": "act", "state": "works", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-xhaka3456-act-openarm" ] }, "inference_real": { "state": "works", "notes": "Documented on the bimanual OpenArm (--robot.type=bi_openarm_follower, three OpenCV cameras) with OpenArm Mini leaders, running the policy under lerobot-rollout --strategy.type=dagger with standard synchronous inference; human interventions are recorded for fine-tuning.", "evidence": [ "ev-lerobot-hil-openarm-default", "ev-lerobot-hil-openarm-mode" ] } }, "evidence": [ { "id": "ev-lerobot-hil-openarm-default", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/hil_data_collection.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot's human-in-the-loop data collection guide runs trained policies on the bimanual OpenArm follower by default.", "quote": "> The provided commands default to `bi_openarm_follower` + `bi_openarm_mini`." }, { "id": "ev-lerobot-hil-openarm-mode", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/hil_data_collection.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "The guide names this policy for the lerobot-rollout dagger command on bi_openarm_follower (standard synchronous inference).", "quote": "**Standard inference (ACT, Diffusion Policy):**" }, { "id": "ev-hub-xhaka3456-act-openarm", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/xhaka3456/act_openarm_pick_red_hat_and_place_on_gray_case", "verified_on": "2026-10-24", "summary": "config.json type=act; train_config.json dataset xhaka3456/openarm_pick_red_hat_and_place_on_gray_case whose meta/info.json robot_type=openarm_bimanual (45 episodes)." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/openarm/act" }, { "robot": "openarm", "policy": "diffusion", "state": "works", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "inference_real": { "state": "works", "notes": "Documented on the bimanual OpenArm (--robot.type=bi_openarm_follower, three OpenCV cameras) with OpenArm Mini leaders, running the policy under lerobot-rollout --strategy.type=dagger with standard synchronous inference; human interventions are recorded for fine-tuning.", "evidence": [ "ev-lerobot-hil-openarm-default", "ev-lerobot-hil-openarm-mode" ] } }, "evidence": [ { "id": "ev-lerobot-hil-openarm-default", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/hil_data_collection.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot's human-in-the-loop data collection guide runs trained policies on the bimanual OpenArm follower by default.", "quote": "> The provided commands default to `bi_openarm_follower` + `bi_openarm_mini`." }, { "id": "ev-lerobot-hil-openarm-mode", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/hil_data_collection.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "The guide names this policy for the lerobot-rollout dagger command on bi_openarm_follower (standard synchronous inference).", "quote": "**Standard inference (ACT, Diffusion Policy):**" } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/openarm/diffusion" }, { "robot": "openarm", "policy": "eo1", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-crislmfroes-eo1-openarm" ] } }, "evidence": [ { "id": "ev-hub-crislmfroes-eo1-openarm", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/crislmfroes/eo1-openarm-lift-fanta-bottle-clean", "verified_on": "2026-10-24", "summary": "config.json type=eo1; train_config.json dataset crislmfroes/openarm-lift-fanta-bottle-clean whose meta/info.json robot_type=openarm_follower (54 episodes); 3000 steps." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/openarm/eo1" }, { "robot": "openarm", "policy": "groot-n1-7", "state": "works-with-caveats", "framework": "isaac-groot", "caveats": [ "Documented only in Enactic's fork of Isaac-GR00T (branch n1.7-release-ga-openarm, a single commit ahead of NVIDIA main as of 2026-10-23), not in NVIDIA's Isaac-GR00T or in LeRobot.", "Bimanual OpenArm only (OPENARM_BIMANUAL_REL tag, 16-D joint actions, both wrist cameras required); robot control runs in Enactic's Dora graph, which talks to the policy server over a UNIX socket.", "Isaac-GR00T's scripts/deployment tools (TensorRT, standalone inference) do not apply OpenArm camera profiles and are not supported for OpenArm checkpoints in this release.", "No on-robot success rates are published; the vendor checkpoint card lists inputs, outputs and training tasks only." ], "capabilities": { "train": { "state": "works", "notes": "Fine-tune nvidia/GR00T-N1.7-3B on OpenArm recordings converted with enactic/openarm_dataset (LeRobot v2.1 + modality.json); vendor checkpoint enactic/GR00T-N1.7-openarm-2-cell-pillow trained on five pillow tasks.", "evidence": [ "ev-enactic-groot-openarm-train", "ev-enactic-groot-openarm-card" ] }, "inference_real": { "state": "works", "notes": "Vendor-documented deployment path: openarm_server policy server (PyTorch, 32-step chunks at 30 Hz) serving dora-openarm-local-policy-server.", "evidence": [ "ev-enactic-groot-openarm-server", "ev-enactic-groot-openarm-card" ] } }, "evidence": [ { "id": "ev-enactic-groot-openarm-train", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/enactic/Isaac-GR00T/blob/a7772f0709dcf2d43227d8a63908af6aea8dfa0e/examples/OpenArm/README.md", "date": "2026-10-23", "verified_on": "2026-10-25", "summary": "Enactic (OpenArm maker) 'OpenArm N1.7 Training' guide: fine-tune GR00T N1.7 on OpenArm relative-joint data with the OPENARM_BIMANUAL_REL embodiment tag.", "quote": "The `OPENARM_BIMANUAL_REL` tag (projector slot 12) outputs 16D joint actions:" }, { "id": "ev-enactic-groot-openarm-server", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/enactic/Isaac-GR00T/blob/a7772f0709dcf2d43227d8a63908af6aea8dfa0e/openarm_server/README.md", "date": "2026-10-23", "verified_on": "2026-10-25", "summary": "Enactic 'OpenArm Policy Server' serves a GR00T N1.7 OpenArm checkpoint to dora-openarm-local-policy-server; robot control, interpolation and recording stay in the Dora graph.", "quote": "Serves a GR00T N1.7 OpenArm checkpoint to" }, { "id": "ev-enactic-groot-openarm-card", "tier": "upstream-documented", "type": "model-card", "url": "https://huggingface.co/enactic/GR00T-N1.7-openarm-2-cell-pillow", "date": "2026-10-06", "verified_on": "2026-10-25", "summary": "Enactic checkpoint card: GR00T N1.7 (base nvidia/GR00T-N1.7-3B) fine-tuned for a bimanual OpenArm pillow task, served with openarm_server.", "quote": "A GR00T N1.7 vision-language-action policy fine-tuned for the bimanual OpenArm" } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/openarm/groot-n1-7" }, { "robot": "openarm", "policy": "molmoact2", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-ahiroakiishikawa-molmoact2-openarm" ] } }, "evidence": [ { "id": "ev-hub-ahiroakiishikawa-molmoact2-openarm", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/aHiroakiIshikawa/molmoact2-openarm-dual-cam-sponge-v4-last", "verified_on": "2026-10-24", "summary": "config.json type=molmoact2; train_config.json dataset aHiroakiIshikawa/openarm_mini_dual_camera_blue_sponge_0529_01_right_only_lpf_hybrid_v4 whose meta/info.json robot_type=openarm_follower (141 episodes); 100000 steps." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/openarm/molmoact2" }, { "robot": "openarm", "policy": "pi0", "state": "works", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "inference_real": { "state": "works", "notes": "Documented on the bimanual OpenArm (--robot.type=bi_openarm_follower, three OpenCV cameras) with OpenArm Mini leaders, running the policy under lerobot-rollout --strategy.type=dagger with RTC inference (--inference.type=rtc); human interventions are recorded for fine-tuning.", "evidence": [ "ev-lerobot-hil-openarm-default", "ev-lerobot-hil-openarm-mode" ] } }, "evidence": [ { "id": "ev-lerobot-hil-openarm-default", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/hil_data_collection.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot's human-in-the-loop data collection guide runs trained policies on the bimanual OpenArm follower by default.", "quote": "> The provided commands default to `bi_openarm_follower` + `bi_openarm_mini`." }, { "id": "ev-lerobot-hil-openarm-mode", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/hil_data_collection.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "The guide names this policy for the lerobot-rollout dagger command on bi_openarm_follower (RTC inference (--inference.type=rtc)).", "quote": "**With RTC for large models (Pi0, Pi0.5, SmolVLA):**" } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/openarm/pi0" }, { "robot": "openarm", "policy": "pi05", "state": "works", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "inference_real": { "state": "works", "notes": "Documented on the bimanual OpenArm (--robot.type=bi_openarm_follower, three OpenCV cameras) with OpenArm Mini leaders, running the policy under lerobot-rollout --strategy.type=dagger with RTC inference (--inference.type=rtc); human interventions are recorded for fine-tuning.", "evidence": [ "ev-lerobot-hil-openarm-default", "ev-lerobot-hil-openarm-mode" ] } }, "evidence": [ { "id": "ev-lerobot-hil-openarm-default", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/hil_data_collection.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot's human-in-the-loop data collection guide runs trained policies on the bimanual OpenArm follower by default.", "quote": "> The provided commands default to `bi_openarm_follower` + `bi_openarm_mini`." }, { "id": "ev-lerobot-hil-openarm-mode", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/hil_data_collection.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "The guide names this policy for the lerobot-rollout dagger command on bi_openarm_follower (RTC inference (--inference.type=rtc)).", "quote": "**With RTC for large models (Pi0, Pi0.5, SmolVLA):**" } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/openarm/pi05" }, { "robot": "openarm", "policy": "smolvla", "state": "works", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-gtgando-smolvla-openarm", "ev-hub-tetraxcode-smolvla-openarm" ] }, "inference_real": { "state": "works", "notes": "Documented on the bimanual OpenArm (--robot.type=bi_openarm_follower, three OpenCV cameras) with OpenArm Mini leaders, running the policy under lerobot-rollout --strategy.type=dagger with RTC inference (--inference.type=rtc); human interventions are recorded for fine-tuning.", "evidence": [ "ev-lerobot-hil-openarm-default", "ev-lerobot-hil-openarm-mode" ] } }, "evidence": [ { "id": "ev-lerobot-hil-openarm-default", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/hil_data_collection.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot's human-in-the-loop data collection guide runs trained policies on the bimanual OpenArm follower by default.", "quote": "> The provided commands default to `bi_openarm_follower` + `bi_openarm_mini`." }, { "id": "ev-lerobot-hil-openarm-mode", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/hil_data_collection.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "The guide names this policy for the lerobot-rollout dagger command on bi_openarm_follower (RTC inference (--inference.type=rtc)).", "quote": "**With RTC for large models (Pi0, Pi0.5, SmolVLA):**" }, { "id": "ev-hub-gtgando-smolvla-openarm", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/gtgando/openarm-smolvla-pick-place-40k", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config.json dataset gtgando/openarm-pick-place-50ep whose meta/info.json robot_type=openarm_follower (100 episodes)." }, { "id": "ev-hub-tetraxcode-smolvla-openarm", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/TetraxCode/my_smolvla_openarm", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config.json dataset TetraxCode/openarm_pick_place_v1 whose meta/info.json robot_type=openarm_bimanual (19 episodes); fine-tuned from lerobot/smolvla_base." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/openarm/smolvla" }, { "robot": "openarm", "policy": "vla-jepa", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-crislmfroes-vlajepa-openarm" ] } }, "evidence": [ { "id": "ev-hub-crislmfroes-vlajepa-openarm", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/crislmfroes/vla-jepa-openarm-lift-fanta-bottle-clean", "verified_on": "2026-10-24", "summary": "config.json type=vla_jepa; train_config.json dataset crislmfroes/openarm-lift-fanta-bottle-clean whose meta/info.json robot_type=openarm_follower (54 episodes); pretrained from lerobot/VLA-JEPA-Pretrain." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/openarm/vla-jepa" }, { "robot": "openarm", "policy": "vqbet", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-louislerikiki-vqbet-wrist", "ev-hub-louislerikiki-vqbet-top" ] } }, "evidence": [ { "id": "ev-hub-louislerikiki-vqbet-wrist", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Louislerikiki/vqbet_sort_by_weight_wrist_v1", "date": "2026-07-08", "verified_on": "2026-10-25", "summary": "config.json type=vqbet; train_config.json dataset Louislerikiki/bimanual_sort_by_weight_v2 (100000 steps), whose meta/info.json robot_type=bi_openarm_follower (100 episodes)." }, { "id": "ev-hub-louislerikiki-vqbet-top", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Louislerikiki/vqbet_sort_by_weight_topview_v1", "date": "2026-07-08", "verified_on": "2026-10-25", "summary": "config.json type=vqbet; same dataset Louislerikiki/bimanual_sort_by_weight_v2 (robot_type=bi_openarm_follower), 100000 steps." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/openarm/vqbet" }, { "robot": "openarm", "policy": "xvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-leapshared-xvla" ] } }, "evidence": [ { "id": "ev-hub-leapshared-xvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/leapshared/Incline_new_60epi", "date": "2026-09-02", "verified_on": "2026-10-25", "summary": "config.json type=xvla (init lerobot/xvla-base); train_config.json dataset leapshared/Incline_new_20260902_203009 (20000 steps), whose meta/info.json robot_type=bi_openarm_follower (60 episodes). Card explains the required camera rename map; no rollout results." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/openarm/xvla" }, { "robot": "reachy2", "policy": "act", "state": "works-with-caveats", "framework": "lerobot", "framework_version_tested": "0.6.1", "caveats": [ "On-robot inference is community-reported only. Upstream documents training and recording; its evaluation step uses lerobot-eval, which in v0.6.1 evaluates in a simulation env, so it does not document on-robot inference.", "The only result is a score of 5 under a 'Performance (10/10)' header in a community README, with no stated protocol; the README does not say explicitly that the score came from physical trials, though its documented ACT deployment targets the physical robot r2-0008.", "Tested configuration: both arms, neck and torso camera at 640x480; mobile base, antennas and teleop cameras disabled. The LeRobot version used is not stated." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-lerobot-reachy2-train", "ev-hub-erlhub-act-reachy2" ] }, "inference_real": { "state": "works-with-caveats", "notes": "Community deployment via lerobot-record --policy.path on a physical Reachy 2; scored 5 on an undefined /10 scale.", "evidence": [ "ev-reachy-pollen-act-deploy", "ev-reachy-pollen-registry" ] }, "teleop_record": { "state": "works", "notes": "Recording uses --robot.type=reachy2 with --teleop.type=reachy2_teleoperator; the page also documents Pollen's own VR data-acquisition app.", "evidence": [ "ev-lerobot-reachy2-record" ] } }, "evidence": [ { "id": "ev-lerobot-reachy2-train", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/reachy2.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "The Reachy 2 page's 'Step 3: Train' runs lerobot-train with --policy.type=act on the recorded Reachy 2 dataset.", "quote": "--policy.type=act \\\n --output_dir=outputs/train/reachy2_test" }, { "id": "ev-lerobot-reachy2-record", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/reachy2.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "Upstream states Reachy 2 is supported by LeRobot's recording features.", "quote": "Reachy 2 is fully supported by LeRobot’s recording features." }, { "id": "ev-lerobot-eval-config", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/src/lerobot/configs/eval.py", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "In v0.6.1, EvalPipelineConfig (the lerobot-eval CLI config) has a required env field and no robot field.", "quote": "env: envs.EnvConfig\n eval: EvalConfig = field(default_factory=EvalConfig)\n policy: PreTrainedConfig | None = None" }, { "id": "ev-hub-erlhub-act-reachy2", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/erl-hub/reachy-act-execute-pick-and-place", "date": "2026-04-13", "verified_on": "2026-10-25", "summary": "config.json type=act; train_config.json dataset erl-hub/reachy-pick-and-place-images whose meta/info.json robot_type=reachy2." }, { "id": "ev-reachy-pollen-act-deploy", "tier": "community-reported", "type": "source-code", "url": "https://github.com/NikolaRaicevic2001/reachy_pollen/blob/ae76a3eedbe07a7a900262b31e986e2e4bec45a2/README.md", "date": "2026-04-14", "verified_on": "2026-10-25", "summary": "Community README's 'Deploying policies on reachy2' section runs lerobot-record with --robot.type=reachy2 on robot r2-0008 at 192.168.10.172 (the same address its VR-teleoperation section uses for the physical robot) and --policy.path=erl-hub/reachy-act-execute-pick-and-place.", "quote": "--policy.path=erl-hub/reachy-act-execute-pick-and-place" }, { "id": "ev-reachy-pollen-registry", "tier": "community-reported", "type": "source-code", "url": "https://github.com/NikolaRaicevic2001/reachy_pollen/blob/7edb3a080d1cd68d88a0f03a518589260e9110dd/README.md", "date": "2026-08-21", "verified_on": "2026-10-25", "summary": "The README's policy registry ('policies you train or deploy') scores the ACT run trained on erl-hub/reachy-pick-and-place-images 5 under a 'Performance (10/10)' header; no scoring protocol is given.", "quote": "| `act` | Action Chunking Transformer (`--policy.type=act`) | 250MB | 2026-04-14_01-02-26-erl-hub-reachy-pick-and-place-images_s7908 | CPU; Intel Core 9 (RTX GeForce 2080+) | 5 |" } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/reachy2/act" }, { "robot": "reachy2", "policy": "pi0", "state": "partial", "partial_scope": "Simulation only: a π0 checkpoint fine-tuned on simulated Reachy 2 teleoperation data (TAVIS benchmark) with a simulation eval script. No real-robot training or inference is documented.", "framework": "lerobot", "caveats": [ "Community-reported only (TAVIS benchmark, anonymous submission under double-blind review); no Reachy 2 success rates are published yet." ], "capabilities": { "train": { "state": "works", "notes": "Simulated (VR-teleoperated) TAVIS-Hands Reachy 2 data, not real-robot data.", "evidence": [ "ev-tavis-pi0-reachy2-checkpoint", "ev-tavis-org-card" ] }, "inference_sim": { "state": "partial", "notes": "The benchmark ships an eval script that rolls out its π0 checkpoints in simulation; no Reachy 2 results are published.", "evidence": [ "ev-tavis-org-card" ] } }, "evidence": [ { "id": "ev-tavis-pi0-reachy2-checkpoint", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/tavis-benchmark/pi0-tavis-hands-reachy2/blob/db7ee5cadd2eda28a2582bfb382032847a4272bb/README.md", "date": "2026-05-04", "verified_on": "2026-10-25", "summary": "config.json type=pi0 (head + two wrist cameras, 36-dim state, 19-dim action), base_model lerobot/pi0_base, card dataset tavis-benchmark/tavis-hands-reachy2 (LeRobotDataset v3.0, robot_type null; robot identified by the card and org page).", "quote": "π₀ multi-task checkpoint, fine-tuned on the TAVIS-Hands suite for the Reachy2 humanoid with head + bimanual wrist cameras." }, { "id": "ev-tavis-org-card", "tier": "community-reported", "type": "community-post", "url": "https://huggingface.co/spaces/tavis-benchmark/README/blob/0674f64be968791a6d75fc237c0d77621369b99c/README.md", "date": "2026-05-31", "verified_on": "2026-10-25", "summary": "TAVIS org card: eight simulated manipulation tasks on Fourier GR1T2 and Pollen Reachy 2, VR-teleoperated demos, π0 baselines; tavis-hands-reachy2 (300 episodes) and pi0-tavis-hands-reachy2 listed as Reachy2.", "quote": "TAVIS ships eight simulated manipulation tasks across two robots (**Fourier GR1T2** and **Pollen Reachy 2**)" } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/reachy2/pi0" }, { "robot": "reachy2", "policy": "pi05", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-erlhub-pi05-reachy2" ] } }, "evidence": [ { "id": "ev-hub-erlhub-pi05-reachy2", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/erl-hub/reachy-pi05", "verified_on": "2026-10-24", "summary": "LoRA checkpoint at 2026-04-30_02-32-39-pi05-lora-ft-erl-hub-reachy-pick-and-place-images_s6246/checkpoints/030000/pretrained_model: config.json type=pi05; train_config.json dataset erl-hub/reachy-pick-and-place-images whose meta/info.json robot_type=reachy2 (25 episodes); pretrained from lerobot/pi05_base." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/reachy2/pi05" }, { "robot": "reachy2", "policy": "smolvla", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "The only result is a score of 6.5 under a 'Performance (10/10)' header in a community README, with no stated protocol; the README does not say explicitly that the score came from physical trials, though its deployment command runs a SmolVLA checkpoint on the physical robot r2-0008.", "The scored run (erl-hub/reachy-smolvla, run s2226) and the checkpoint in the deployment command (erl-hub/reachy-smolvla-ft-execute-pick-and-place) are different Hub repos.", "Tested configuration: both arms, neck and torso camera at 640x480 (SmolVLA's other two camera slots empty); mobile base, antennas and teleop cameras disabled. The LeRobot version used is not stated." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-erlhub-smolvla-reachy2", "ev-hub-erlhub-smolvla-ft-reachy2" ] }, "inference_real": { "state": "works-with-caveats", "notes": "Community deployment via lerobot-record --policy.path on a physical Reachy 2; scored 6.5 on an undefined /10 scale.", "evidence": [ "ev-reachy-pollen-smolvla-deploy" ] } }, "evidence": [ { "id": "ev-hub-erlhub-smolvla-reachy2", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/erl-hub/reachy-smolvla", "verified_on": "2026-10-24", "summary": "Checkpoint at 2026-04-25_05-03-29-smolvla-basic-run-erl-hub-reachy-pick-and-place-images_s2226/checkpoints/020000/pretrained_model: config.json type=smolvla; train_config.json dataset erl-hub/reachy-pick-and-place-images whose meta/info.json robot_type=reachy2 (25 episodes)." }, { "id": "ev-hub-erlhub-smolvla-ft-reachy2", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/erl-hub/reachy-smolvla-ft-execute-pick-and-place", "date": "2026-04-28", "verified_on": "2026-10-25", "summary": "config.json type=smolvla; train_config.json dataset erl-hub/reachy-pick-and-place-images (robot_type=reachy2). This is the checkpoint named in the community deployment command." }, { "id": "ev-reachy-pollen-smolvla-deploy", "tier": "community-reported", "type": "source-code", "url": "https://github.com/NikolaRaicevic2001/reachy_pollen/blob/7edb3a080d1cd68d88a0f03a518589260e9110dd/README.md", "date": "2026-08-21", "verified_on": "2026-10-25", "summary": "Community README: 'Deploying policies on reachy2' runs lerobot-record with --robot.type=reachy2 on r2-0008 at 192.168.137.100 (the on-robot 'bedrock' host of its 'Recording: On the Robot' section) and --policy.path=erl-hub/reachy-smolvla-ft-execute-pick-and-place; its policy registry scores SmolVLA run s2226 at 6.5 under 'Performance (10/10)'.", "quote": "| `SmolVla` | VLA (`--policy.type=SmolVla`) | 1GB | 2026-04-25_05-03-29-smolvla-basic-run-erl-hub-reachy-pick-and-place-images_s2226 | NVIDIA-L40 CPU 3.8-4.0 | 6.5 |" } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/reachy2/smolvla" }, { "robot": "rebot-b601", "policy": "act", "state": "partial", "partial_scope": "Training on reBot B601-DM data is documented by the vendor (Seeed wiki 'B601-DM with LeRobot'). The wiki's on-robot evaluation command (lerobot-record with --policy.path and no teleoperator) is rejected by the Seeed LeRobot fork the wiki installs, which directs policy deployment to lerobot-rollout, so on-robot inference is not validly documented. Public checkpoints trained on B601-DM data also exist.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-seeed-wiki-b601-dm-lerobot", "ev-hub-nikodembartnik-act-rebot", "ev-hub-lakesenberg-act-rebot" ] } }, "evidence": [ { "id": "ev-seeed-wiki-b601-dm-lerobot", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/Seeed-Studio/wiki-documents/blob/86afe5634df5713aec98cf77c09a7e8d670feb1d/sites/en/docs/Robotics/Robot_Kits/reBot_Arm/B601_DM/reBot_Arm_B601_DM_Lerobot.md", "date": "2026-10-15", "verified_on": "2026-10-25", "summary": "Seeed wiki 'B601-DM with LeRobot' (Training and Evaluation section). ACT tab: lerobot-train with --policy.type=act on a reBot dataset (300k steps), followed by an evaluation via lerobot-record.", "quote": "**Policy type**: We provide the policy with `policy.type=act`." }, { "id": "ev-seeed-lerobot-fork-record", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/Seeed-Projects/lerobot/blob/4aaff99be4a1d81568c08c8f0296b41b40c99ec4/src/lerobot/scripts/lerobot_record.py", "date": "2026-08-28", "verified_on": "2026-10-25", "summary": "Seeed's LeRobot fork (main, pyproject version 0.6.2), which the wiki tells users to install: RecordConfig raises an error when no teleoperator is given and points policy deployment to lerobot-rollout.", "quote": "For policy-based deployment, use lerobot-rollout instead." }, { "id": "ev-hub-nikodembartnik-act-rebot", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/nikodembartnik/act_rebot_cube", "verified_on": "2026-10-24", "summary": "config.json type=act; train_config.json dataset nikodembartnik/rebot_cube whose meta/info.json robot_type=rebot_b601_follower (LeRobot-native B601-DM, 50 episodes)." }, { "id": "ev-hub-lakesenberg-act-rebot", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Lakesenberg/act_rebot_mian_model", "verified_on": "2026-10-24", "summary": "config.json type=act; train_config.json dataset Lakesenberg/Seeed_Hackathon_noodels whose meta/info.json robot_type=seeed_b601_dm_follower (Seeed plugin naming for the B601-DM, 17 episodes)." } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/rebot-b601/act" }, { "robot": "rebot-b601", "policy": "groot-n1-7", "state": "works-with-caveats", "caveats": [ "Documented in NVIDIA's Isaac-GR00T repository, not through LeRobot's groot policy: the real-robot client uses Seeed's LeRobot fork plus the lerobot-robot-seeed-b601 plugin (--robot.type=seeed_b601_dm_follower), not LeRobot's native rebot_b601_follower.", "Covers the reBot Arm B601-DM (Damiao-motor variant); datasets must be converted from LeRobot v3 to v2 before Isaac-GR00T fine-tuning.", "The example landed on Isaac-GR00T main (2026-07-30) after the n1.7-release tag; it is not part of a tagged release." ], "framework": "isaac-groot", "capabilities": { "train": { "state": "works", "notes": "Fine-tuned from nvidia/GR00T-N1.7-3B with --embodiment-tag NEW_EMBODIMENT; six arm joints as relative actions, gripper absolute.", "evidence": [ "ev-gr00t-rebot" ] }, "inference_real": { "state": "works", "notes": "Closed-loop deployment via run_gr00t_server.py plus the eval_rebot_arm_dm.py client; the README shows only open-loop plots, no closed-loop success rates.", "evidence": [ "ev-gr00t-rebot", "ev-gr00t-rebot-client" ] } }, "evidence": [ { "id": "ev-gr00t-rebot", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/NVIDIA/Isaac-GR00T/blob/b9955401d50c92a29258732e3ad6ccd579f1bdc0/examples/rebot-arm-dm/README.md", "date": "2026-07-30", "verified_on": "2026-10-24", "summary": "NVIDIA's Isaac-GR00T example fine-tunes GR00T N1.7 on reBot Arm B601-DM data and evaluates it on the real robot.", "quote": "This guide shows how to finetune dataset collected from [rebot arm b601 dm](https://www.seeedstudio.com/reBot-Arm-B601-DM-Bundle.html) robot, and evaluate the model on the real robot." }, { "id": "ev-gr00t-rebot-client", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/NVIDIA/Isaac-GR00T/blob/b9955401d50c92a29258732e3ad6ccd579f1bdc0/examples/rebot-arm-dm/pyproject.toml", "date": "2026-07-30", "verified_on": "2026-10-24", "summary": "The deployment client depends on Seeed's LeRobot fork and the lerobot-robot-seeed-b601 robot plugin at pinned commits.", "quote": "\"lerobot-robot-seeed-b601 @ git+https://github.com/Seeed-Projects/lerobot-robot-seeed-b601.git@5d36bef6e352ddfc4cefc40a16db02772e25655c\"," } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/rebot-b601/groot-n1-7" }, { "robot": "rebot-b601", "policy": "molmoact2", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-harshmuriki-molmoact2-rebot", "ev-hub-lilkm-molmoact2-birebot" ] } }, "evidence": [ { "id": "ev-hub-harshmuriki-molmoact2-rebot", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Harshmuriki/molmoact2_finetuned_seeed_rebot_100k", "verified_on": "2026-10-24", "summary": "config.json type=molmoact2; train_config.json dataset Harshmuriki/rebot_pick_place_wrist_v3 whose meta/info.json robot_type=seeed_b601_dm_follower (Seeed plugin naming for the B601-DM, 100 episodes)." }, { "id": "ev-hub-lilkm-molmoact2-birebot", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/lilkm/molmoact2_bi_rebot", "verified_on": "2026-10-24", "summary": "config.json type=molmoact2; train_config.json dataset lilkm/bi_rebot_pick_and_place_cubes whose meta/info.json robot_type=bi_rebot_b601_follower (bimanual B601-DM, 50 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/rebot-b601/molmoact2" }, { "robot": "rebot-b601", "policy": "multitask-dit", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-sielo-dit" ] } }, "evidence": [ { "id": "ev-hub-sielo-dit", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/SieloRobotics/dit_ee_cubes", "date": "2026-10-20", "verified_on": "2026-10-25", "summary": "config.json type=multi_task_dit; train_config.json dataset SieloRobotics/rebot_pick_place_ee (40000 steps), whose meta/info.json robot_type=rebot_b601_follower (512 episodes). Card has no rollout results." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/rebot-b601/multitask-dit" }, { "robot": "rebot-b601", "policy": "pi0", "state": "partial", "partial_scope": "Training on reBot B601-DM data is documented by the vendor (Seeed wiki 'B601-DM with LeRobot'). The wiki's on-robot evaluation command (lerobot-record with --policy.path and no teleoperator) is rejected by the Seeed LeRobot fork the wiki installs, which directs policy deployment to lerobot-rollout, so on-robot inference is not validly documented.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-seeed-wiki-b601-dm-lerobot" ] } }, "evidence": [ { "id": "ev-seeed-wiki-b601-dm-lerobot", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/Seeed-Studio/wiki-documents/blob/86afe5634df5713aec98cf77c09a7e8d670feb1d/sites/en/docs/Robotics/Robot_Kits/reBot_Arm/B601_DM/reBot_Arm_B601_DM_Lerobot.md", "date": "2026-10-15", "verified_on": "2026-10-25", "summary": "Seeed wiki 'B601-DM with LeRobot' (Training and Evaluation section). Pi0 tab: lerobot-train --policy.type=pi0 --policy.pretrained_path=lerobot/pi0_base on your reBot dataset, then evaluate with lerobot-record --robot.type=seeed_b601_dm_follower --policy.path=...", "quote": "Using it in LeRobot is simple: just set the policy type to `--policy.type=pi0` during training (no need to repeat the general training/evaluation concepts covered in the ACT section)." }, { "id": "ev-seeed-lerobot-fork-record", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/Seeed-Projects/lerobot/blob/4aaff99be4a1d81568c08c8f0296b41b40c99ec4/src/lerobot/scripts/lerobot_record.py", "date": "2026-08-28", "verified_on": "2026-10-25", "summary": "Seeed's LeRobot fork (main, pyproject version 0.6.2), which the wiki tells users to install: RecordConfig raises an error when no teleoperator is given and points policy deployment to lerobot-rollout.", "quote": "For policy-based deployment, use lerobot-rollout instead." } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/rebot-b601/pi0" }, { "robot": "rebot-b601", "policy": "pi05", "state": "partial", "partial_scope": "Training on reBot B601-DM data is documented by the vendor (Seeed wiki 'B601-DM with LeRobot'). The wiki's on-robot evaluation command (lerobot-record with --policy.path and no teleoperator) is rejected by the Seeed LeRobot fork the wiki installs, which directs policy deployment to lerobot-rollout, so on-robot inference is not validly documented. Public checkpoints trained on B601-DM data also exist.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-seeed-wiki-b601-dm-lerobot", "ev-hub-tinjyuu-pi05-rebot", "ev-hub-qualia-pi05-rebot" ] } }, "evidence": [ { "id": "ev-seeed-wiki-b601-dm-lerobot", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/Seeed-Studio/wiki-documents/blob/86afe5634df5713aec98cf77c09a7e8d670feb1d/sites/en/docs/Robotics/Robot_Kits/reBot_Arm/B601_DM/reBot_Arm_B601_DM_Lerobot.md", "date": "2026-10-15", "verified_on": "2026-10-25", "summary": "Seeed wiki 'B601-DM with LeRobot' (Training and Evaluation section). Pi0.5 tab: lerobot-train --policy.type=pi05 --policy.pretrained_path=lerobot/pi05_base on your reBot dataset, then evaluate with lerobot-record --robot.type=seeed_b601_dm_follower --policy.path=...", "quote": "To use it in LeRobot: just set the policy type to `--policy.type=pi05`." }, { "id": "ev-seeed-lerobot-fork-record", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/Seeed-Projects/lerobot/blob/4aaff99be4a1d81568c08c8f0296b41b40c99ec4/src/lerobot/scripts/lerobot_record.py", "date": "2026-08-28", "verified_on": "2026-10-25", "summary": "Seeed's LeRobot fork (main, pyproject version 0.6.2), which the wiki tells users to install: RecordConfig raises an error when no teleoperator is given and points policy deployment to lerobot-rollout.", "quote": "For policy-based deployment, use lerobot-rollout instead." }, { "id": "ev-hub-tinjyuu-pi05-rebot", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/tinjyuu/pi05-rebot-prod1-peft-15k", "verified_on": "2026-10-24", "summary": "config.json type=pi05; train_config.json dataset tinjyuu/rebot-prod1 whose meta/info.json robot_type=seeed_b601_dm_follower (Seeed plugin naming for the B601-DM, 43 episodes); pretrained from lerobot/pi05_base." }, { "id": "ev-hub-qualia-pi05-rebot", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/qualia-robotics/pi05-b601-dm-icehockey-v1-72e6d068", "verified_on": "2026-10-24", "summary": "config.json type=pi05; train_config.json dataset userresu2026/b601-dm-icehockey-v1 whose meta/info.json robot_type=seeed_b601_dm_follower (125 episodes)." } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/rebot-b601/pi05" }, { "robot": "rebot-b601", "policy": "smolvla", "state": "partial", "partial_scope": "Training on reBot B601-DM data is documented by the vendor (Seeed wiki 'B601-DM with LeRobot'). The wiki's on-robot evaluation command (lerobot-record with --policy.path and no teleoperator) is rejected by the Seeed LeRobot fork the wiki installs, which directs policy deployment to lerobot-rollout, so on-robot inference is not validly documented. Public checkpoints trained on B601-DM data also exist.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-seeed-wiki-b601-dm-lerobot", "ev-hub-namin72-smolvla-rebot" ] } }, "evidence": [ { "id": "ev-seeed-wiki-b601-dm-lerobot", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/Seeed-Studio/wiki-documents/blob/86afe5634df5713aec98cf77c09a7e8d670feb1d/sites/en/docs/Robotics/Robot_Kits/reBot_Arm/B601_DM/reBot_Arm_B601_DM_Lerobot.md", "date": "2026-10-15", "verified_on": "2026-10-25", "summary": "Seeed wiki 'B601-DM with LeRobot' (Training and Evaluation section). SmolVLA tab: fine-tune lerobot/smolvla_base on your reBot dataset (lerobot-train, 20k steps), then evaluate with lerobot-record --robot.type=seeed_b601_dm_follower --policy.path=...", "quote": "Use `smolvla_base` (the pretrained 450M model) as a starting point and fine-tune on your dataset." }, { "id": "ev-seeed-lerobot-fork-record", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/Seeed-Projects/lerobot/blob/4aaff99be4a1d81568c08c8f0296b41b40c99ec4/src/lerobot/scripts/lerobot_record.py", "date": "2026-08-28", "verified_on": "2026-10-25", "summary": "Seeed's LeRobot fork (main, pyproject version 0.6.2), which the wiki tells users to install: RecordConfig raises an error when no teleoperator is given and points policy deployment to lerobot-rollout.", "quote": "For policy-based deployment, use lerobot-rollout instead." }, { "id": "ev-hub-namin72-smolvla-rebot", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/namin72/smolvla-rebot-b601-green-eraser-brown-gray-100k", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config.json dataset namin72/pick-up-green-eraser-place-in-brown-gray-baskets-rgbd whose meta/info.json robot_type=rebot_b601_follower (LeRobot-native B601-DM, 192 episodes); fine-tuned from lerobot/smolvla_base." } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/rebot-b601/smolvla" }, { "robot": "rebot-b601", "policy": "vla-jepa", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-nikodembartnik-vlajepa-rebot" ] } }, "evidence": [ { "id": "ev-hub-nikodembartnik-vlajepa-rebot", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/nikodembartnik/vla_jepa_rebot_towel_folding_2026-10-12_14-10-32", "verified_on": "2026-10-24", "summary": "config.json type=vla_jepa; train_config.json dataset nikodembartnik/rebot_mini_towel_folding_merged whose meta/info.json robot_type=rebot_b601_follower (LeRobot-native B601-DM, 200 episodes); pretrained from lerobot/VLA-JEPA-Pretrain." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/rebot-b601/vla-jepa" }, { "robot": "so100", "policy": "act", "state": "works", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-smolvla-paper-so100" ] }, "inference_real": { "state": "works", "evidence": [ "ev-smolvla-paper-so100" ] }, "teleop_record": { "state": "works", "evidence": [ "ev-smolvla-paper-datasets" ] } }, "evidence": [ { "id": "ev-smolvla-paper-so100", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/html/2506.01844v1", "date": "2025-06-02", "verified_on": "2026-10-24", "summary": "SmolVLA paper (Hugging Face/LeRobot team), Table 3: real-world SO100 benchmark (pick-place, stacking, sorting) comparing ACT (single-task), pi0 3.5B and SmolVLA (multi-task).", "quote": "Table 3: Real-world benchmarks (SO100). Success rate (%) across three tasks using policies trained in multi-task and single-task settings." }, { "id": "ev-smolvla-paper-datasets", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/html/2506.01844v1", "date": "2025-06-02", "verified_on": "2026-10-24", "summary": "SmolVLA paper: real-world evaluation datasets were recorded on SO-100 (three tasks) and SO-101 (one task).", "quote": "For real-world evaluation, we collected three datasets using the SO-100 robot arm" } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so100/act" }, { "robot": "so100", "policy": "diffusion", "state": "works-with-caveats", "framework": "lerobot", "framework_version_tested": "0.6.0", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "The physical arms are SO-100, but the run used --robot.type=so101_follower (LeLab uses that type for both arm versions); reproduce with the same driver type.", "One community setup (three 640x480 cameras, one pick-and-place task): 9/10 at trained start positions, 2/2 between them, 0/2 outside the marked area." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-hcltech-diffusion-so100", "ev-hub-neilkim-diffusion-so100", "ev-van-i-so100-stand" ] }, "inference_real": { "state": "works-with-caveats", "notes": "9/10 at trained positions, 0/2 outside the demonstrated area; ~16 s to first arm movement.", "evidence": [ "ev-van-i-so100-stand" ] } }, "evidence": [ { "id": "ev-hub-hcltech-diffusion-so100", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/hcltech-robotics/diffusion_so100_ducktape_3cam", "date": "2025-08-12", "verified_on": "2026-10-24", "summary": "config.json type=diffusion; train_config.json dataset.repo_id=targabor/so100_ducktape_3cam, whose meta/info.json robot_type=so100_follower (50 episodes)." }, { "id": "ev-hub-neilkim-diffusion-so100", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/NeilKim/diffusion-so100-policy-2", "date": "2025-07-02", "verified_on": "2026-10-24", "summary": "config.json type=diffusion; train_config.json dataset.repo_id=NeilKim/so100-teleop-record-2, whose meta/info.json robot_type=so100_follower (61 episodes)." }, { "id": "ev-van-i-so100-stand", "tier": "community-reported", "type": "model-card", "url": "https://huggingface.co/van-i/so100-imitation-learning-stand/blob/c68846b72dea1e43f63a485684dc1a7d3e3d2ade/README.md", "date": "2026-10-04", "verified_on": "2026-10-25", "versions": { "lerobot": "0.6.0" }, "summary": "Write-up of a self-built SO-100 leader/follower stand (LeRobot 0.6.0 via LeLab, RTX 3090): Diffusion Policy trained from scratch for 36,000 steps and tested on the physical arm, 9/10 at trained positions and 0/2 outside the marked area. States so101_follower is the LeLab driver type for both arm versions.", "quote": "SO-100 leader (moved by hand) + SO-100 follower (does the task), built from parts" } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so100/diffusion" }, { "robot": "so100", "policy": "eo1", "state": "works", "framework": "eo1", "capabilities": { "train": { "state": "works", "evidence": [ "ev-eo1-paper-so100" ] }, "inference_real": { "state": "works", "evidence": [ "ev-eo1-paper-so100" ] } }, "evidence": [ { "id": "ev-eo1-paper-so100", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/html/2508.21112v5", "date": "2026-02-25", "verified_on": "2026-10-24", "summary": "EO-1 paper (v5): EO-1 is evaluated in the real world on a LeRobot SO100 among other robots (EO-1's own codebase); SO100-Community data is also in its pretraining mix.", "quote": "We evaluate its performance on both short-horizon and long-horizon tasks using different robots, including Franka Panda, WidowX 250 S, Agibot G-1, and Lerobot SO100." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so100/eo1" }, { "robot": "so100", "policy": "evo1", "state": "works", "framework": "evo-1", "capabilities": { "train": { "state": "works", "notes": "Also trainable through LeRobot's native evo1 policy (community checkpoint).", "evidence": [ "ev-evo1-paper-so100", "ev-hub-beilinghamburger-evo1-so100" ] }, "inference_real": { "state": "works", "evidence": [ "ev-evo1-paper-so100", "ev-evo1-readme-so100-so101" ] } }, "evidence": [ { "id": "ev-evo1-paper-so100", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/html/2511.04555v2", "date": "2025-12-05", "verified_on": "2026-10-24", "summary": "Evo-1 paper (v2), appendix 'Deployment on LeRobot SO-100 robot': Evo-1 deployed on an SO-100 for a real-world cube pick-and-place task.", "quote": "We deploy our policy on the LeRobot SO-100 platform, a compact desktop manipulator designed for fine-grained tabletop manipulation." }, { "id": "ev-evo1-readme-so100-so101", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/MINT-SJTU/Evo-1/blob/5fd14b015013c4fd0aacf5f8f48f868ca9b870a2/README.md", "date": "2026-08-18", "verified_on": "2026-10-24", "summary": "Official Evo-1 repo README: SO100/SO101 inference via the LeRobot framework (instructions on its evo1-lerobot branch).", "quote": "Added Evo-1 inference in the LeRobot framework for SO100/SO101" }, { "id": "ev-hub-beilinghamburger-evo1-so100", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Beilinghamburger/evo1_so100_vla", "date": "2025-12-11", "verified_on": "2026-10-24", "summary": "config.json type=evo1; train_config.json dataset.repo_id=Beilinghamburger/so100_vla_dataset, whose meta/info.json robot_type=so100_follower (100 episodes). Trained with LeRobot's native evo1 policy." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so100/evo1" }, { "robot": "so100", "policy": "groot-n1-7", "state": "works", "framework": "isaac-groot", "capabilities": { "inference_real": { "state": "works", "notes": "Isaac-GR00T's SO100/SO101 eval script drives a LeRobot so100_follower or so101_follower; the worked example data is SO-101.", "evidence": [ "ev-isaac-groot-eval-so100" ] }, "train": { "state": "works", "notes": "Fine-tuning example uses an so101_follower dataset; SO-100 data follows the same NEW_EMBODIMENT recipe.", "evidence": [ "ev-isaac-groot-so100-example" ] } }, "evidence": [ { "id": "ev-isaac-groot-eval-so100", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/NVIDIA/Isaac-GR00T/blob/n1.7-release/gr00t/eval/real_robot/SO100/eval_so100.py", "date": "2026-04-18", "verified_on": "2026-10-24", "versions": { "isaac-groot": "n1.7-release" }, "summary": "Isaac-GR00T n1.7-release real-robot eval script (LeRobot RobotConfig + GR00T policy server) states it targets both SO100 and SO101; it imports LeRobot so100_follower and so101_follower configs.", "quote": "This script runs closed-loop policy evaluation on the SO100 / SO101 robots\nusing the GR00T Policy API." }, { "id": "ev-isaac-groot-so100-example", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/NVIDIA/Isaac-GR00T/blob/n1.7-release/examples/SO100/README.md", "date": "2026-04-18", "verified_on": "2026-10-24", "versions": { "isaac-groot": "n1.7-release" }, "summary": "Isaac-GR00T n1.7-release 'SO100' example fine-tunes GR00T-N1.7-3B (NEW_EMBODIMENT) on izuluaga/finish_sandwich (meta/info.json robot_type=so101_follower) and runs closed-loop eval with --robot.type=so101_follower.", "quote": "This guide shows how to finetune dataset collected from [SO100](https://huggingface.co/docs/lerobot/en/so101) robot, and evaluate the model on the real robot." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so100/groot-n1-7" }, { "robot": "so100", "policy": "hil-serl", "state": "works-with-caveats", "framework": "lerobot", "framework_version_tested": "0.6.1", "caveats": [ "The guide's documented intervention path is a gamepad or keyboard in end-effector space (needs a URDF); using an SO leader arm as the intervention device is reported broken in open issue #2952, not yet confirmed or fixed by maintainers.", "src/lerobot/configs/train_config_hilserl_so100.json is referenced by the guide but is not in the v0.6.1 source tree; the example configs are hosted in the aractingi/lerobot-example-config-files dataset, and its env_config_so100.json still names the removed so100_follower_end_effector robot type." ], "capabilities": { "train": { "state": "works", "notes": "Online RL on the physical robot (actor/learner).", "evidence": [ "ev-lerobot-hilserl-learner" ] }, "inference_real": { "state": "works", "evidence": [ "ev-lerobot-hilserl-learner" ] }, "teleop_record": { "state": "works-with-caveats", "notes": "Gamepad/keyboard documented; SO leader as intervention device reported broken (#2952).", "evidence": [ "ev-lerobot-hilserl-doc", "ev-gh-lerobot-2952" ] } }, "evidence": [ { "id": "ev-lerobot-hilserl-doc", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/hilserl.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 HIL-SERL real-robot guide: find EE bounds and collect demos on so100_follower/so100_leader, then run the actor/learner with train_config_hilserl_so100.json.", "quote": "lerobot-find-joint-limits \\\n --robot.type=so100_follower \\" }, { "id": "ev-lerobot-hilserl-learner", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/hilserl.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 HIL-SERL guide launches the learner and actor with an SO100 HIL-SERL training config (on-robot RL).", "quote": "python -m lerobot.rl.learner --config_path src/lerobot/configs/train_config_hilserl_so100.json" }, { "id": "ev-gh-lerobot-2952", "tier": "community-reported", "type": "github-issue", "url": "https://github.com/huggingface/lerobot/issues/2952", "date": "2026-02-10", "verified_on": "2026-10-24", "versions": { "lerobot": "0.4.4" }, "summary": "Open issue (LeRobot 0.4.4, labelled bug): the SO leader arm cannot be used as the HIL-SERL intervention device; a 2026-08-02 comment against main reports control_mode is never read by the RL pipeline.", "quote": "TypeError: Teleoperator SOLeader must implement get_teleop_events() method. Compatible teleoperators: GamepadTeleop, KeyboardEndEffectorTeleop" } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so100/hil-serl" }, { "robot": "so100", "policy": "molmoact2", "state": "works-with-caveats", "framework": "lerobot", "framework_version_tested": "0.6.1", "caveats": [ "The MolmoAct2-SO100_101 weights expect the pre-LeRobot-0.5.0 joint calibration convention: lerobot/MolmoAct2-SO100_101-LeRobot includes the frame correction, but self-converted or fine-tuned checkpoints must set joint_signs/joint_offsets or the arm may move in the wrong direction.", "Camera keys must be renamed to cam0/cam1 (--rename_map) to match the checkpoint." ], "capabilities": { "inference_real": { "state": "works-with-caveats", "evidence": [ "ev-lerobot-molmoact2-doc", "ev-lerobot-molmoact2-calib", "ev-molmoact2-so100-101-card" ] }, "train": { "state": "works-with-caveats", "notes": "Fine-tuning from the SO-100/101 checkpoint needs the same joint frame transform.", "evidence": [ "ev-lerobot-molmoact2-calib", "ev-molmoact2-so100-101-card" ] } }, "evidence": [ { "id": "ev-lerobot-molmoact2-doc", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/molmoact2.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 MolmoAct2 guide: zero-shot rollout of lerobot/MolmoAct2-SO100_101-LeRobot on so100_follower (top+side cameras renamed to cam0/cam1).", "quote": "Example with an SO-100 robot using top and side cameras:" }, { "id": "ev-lerobot-molmoact2-calib", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/molmoact2.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 MolmoAct2 guide warns that the SO-100/101 checkpoint uses the pre-0.5.0 joint calibration convention; the converted LeRobot checkpoint includes the frame correction, self-converted/fine-tuned ones need joint_signs/joint_offsets.", "quote": "The MolmoAct2-SO100_101 checkpoint was trained on data that uses a different\njoint calibration convention than LeRobot >= 0.5.0. Without a frame\ncorrection, the arm may move in the wrong direction." }, { "id": "ev-molmoact2-so100-101-card", "tier": "upstream-documented", "type": "model-card", "url": "https://huggingface.co/allenai/MolmoAct2-SO100_101/blob/152569fe57914d97be91055800035f54e250d009/README.md", "date": "2026-05-23", "verified_on": "2026-10-24", "summary": "Ai2's official MolmoAct2-SO100_101 model card: fine-tuned on an SO-100/101 data mixture, intended for SO-100/101 inference and further fine-tuning.", "quote": "This checkpoint is fine-tuned on the SO-100/101 mixture with absolute joint-pose control and annotated language instructions. It is intended for both further fine-tuning and SO-100/101 policy inference." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so100/molmoact2" }, { "robot": "so100", "policy": "pi0-fast", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-maharishiva-pi0fast-so100" ] } }, "evidence": [ { "id": "ev-hub-maharishiva-pi0fast-so100", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/maharishiva/pi0fast_tictactoe09", "verified_on": "2026-10-24", "summary": "config.json type=pi0fast; train_config.json dataset.repo_id=maharishiva/tictactoe-fixed-final, whose meta/info.json robot_type=so100_follower (40 episodes). Uses the earlier 'pi0fast' policy type (pre-'pi0_fast' LeRobot implementation)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so100/pi0-fast" }, { "robot": "so100", "policy": "pi0", "state": "works", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "train": { "state": "works", "notes": "pi0 (3.5B) multi-task baseline trained on the SO100 tasks in the SmolVLA paper.", "evidence": [ "ev-smolvla-paper-so100" ] }, "inference_real": { "state": "works", "evidence": [ "ev-lerobot-inference-pi0-rtc", "ev-smolvla-paper-so100" ] } }, "evidence": [ { "id": "ev-lerobot-inference-pi0-rtc", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/inference.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 rollout guide: Real-Time Chunking example deploys a pi0 policy on so100_follower.", "quote": " --policy.path=${HF_USER}/pi0_policy \\\n --robot.type=so100_follower \\" }, { "id": "ev-smolvla-paper-so100", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/html/2506.01844v1", "date": "2025-06-02", "verified_on": "2026-10-24", "summary": "SmolVLA paper (Hugging Face/LeRobot team), Table 3: real-world SO100 benchmark (pick-place, stacking, sorting) comparing ACT (single-task), pi0 3.5B and SmolVLA (multi-task).", "quote": "Table 3: Real-world benchmarks (SO100). Success rate (%) across three tasks using policies trained in multi-task and single-task settings." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so100/pi0" }, { "robot": "so100", "policy": "pi05", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-so100-pi05-new60" ] } }, "evidence": [ { "id": "ev-hub-so100-pi05-new60", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/tonghuiwang123/so100-pi05-new60/blob/caaeca3a0744222f4e7916c94862eba706b3904f/README.md", "date": "2026-08-29", "verified_on": "2026-10-25", "summary": "config.json type=pi05; train_config.json dataset.repo_id=tonghuiwang123/40 (76000 steps) whose meta/info.json robot_type=so_follower; card names SO-ARM100 / SO100 and gives an lerobot-record --robot.type=so100_follower deploy command (no rollout shown).", "quote": "Pi0.5 policy for **SO-ARM100 / SO100** — `Grab the white cube to the white cup`" } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so100/pi05" }, { "robot": "so100", "policy": "smolvla", "state": "works", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "train": { "state": "works", "evidence": [ "ev-smolvla-paper-so100" ] }, "inference_real": { "state": "works", "evidence": [ "ev-smolvla-paper-so100", "ev-lerobot-async-doc" ] }, "teleop_record": { "state": "works", "evidence": [ "ev-smolvla-paper-datasets" ] } }, "evidence": [ { "id": "ev-smolvla-paper-so100", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/html/2506.01844v1", "date": "2025-06-02", "verified_on": "2026-10-24", "summary": "SmolVLA paper (Hugging Face/LeRobot team), Table 3: real-world SO100 benchmark (pick-place, stacking, sorting) comparing ACT (single-task), pi0 3.5B and SmolVLA (multi-task).", "quote": "Table 3: Real-world benchmarks (SO100). Success rate (%) across three tasks using policies trained in multi-task and single-task settings." }, { "id": "ev-smolvla-paper-datasets", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/html/2506.01844v1", "date": "2025-06-02", "verified_on": "2026-10-24", "summary": "SmolVLA paper: real-world evaluation datasets were recorded on SO-100 (three tasks) and SO-101 (one task).", "quote": "For real-world evaluation, we collected three datasets using the SO-100 robot arm" }, { "id": "ev-lerobot-async-doc", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/async.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 async-inference guide: API example drives an SO100 follower (SO100FollowerConfig) from a SmolVLA policy server.", "quote": "robot_cfg = SO100FollowerConfig(\n port=\"/dev/tty.usbmodem585A0076841\",\n id=\"follower_so100\",\n cameras=camera_cfg\n)" } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so100/smolvla" }, { "robot": "so100", "policy": "wall-oss", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-sosolaris-wallx-so100" ] } }, "evidence": [ { "id": "ev-hub-sosolaris-wallx-so100", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/SoSolaris/wallx_socks_clean", "date": "2026-02-04", "verified_on": "2026-10-24", "summary": "config.json type=wall_x; train_config.json dataset.repo_id=SoSolaris/socks_clean, whose meta/info.json robot_type=so100_follower (200 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so100/wall-oss" }, { "robot": "so100", "policy": "xvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-mcbuehler-xvla-so100", "ev-hub-beilinghamburger-xvla-so100" ] } }, "evidence": [ { "id": "ev-hub-mcbuehler-xvla-so100", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/mcbuehler/xvla_so100_pp_3cams_black_v0", "date": "2026-01-02", "verified_on": "2026-10-24", "summary": "config.json type=xvla; train_config.json dataset.repo_id=mcbuehler/pp_3cams_black_v0, whose meta/info.json robot_type=so100_follower (51 episodes)." }, { "id": "ev-hub-beilinghamburger-xvla-so100", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Beilinghamburger/xvla_so100_vla_200", "date": "2025-12-16", "verified_on": "2026-10-24", "summary": "config.json type=xvla; train_config.json dataset.repo_id=Beilinghamburger/so100_200, whose meta/info.json robot_type=so100_follower (200 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so100/xvla" }, { "robot": "so101", "policy": "act", "state": "works", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "train": { "state": "works", "evidence": [ "ev-lerobot-il-robots-train", "ev-smolvla-paper-so101" ] }, "inference_real": { "state": "works", "evidence": [ "ev-lerobot-act-doc", "ev-smolvla-paper-so101" ] }, "teleop_record": { "state": "works", "evidence": [ "ev-lerobot-il-robots-teleop" ] } }, "evidence": [ { "id": "ev-lerobot-act-doc", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/act.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 ACT guide evaluates a trained ACT policy on the real robot with lerobot-rollout --robot.type=so101_follower.", "quote": "lerobot-rollout \\\n --strategy.type=base \\\n --policy.path=${HF_USER}/act_policy \\\n --robot.type=so101_follower \\" }, { "id": "ev-lerobot-il-robots-train", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/il_robots.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "The v0.6.1 real-robot imitation-learning tutorial records on so101_follower and trains ACT (--policy.type=act) on the resulting so101_test dataset.", "quote": "lerobot-train \\\n --dataset.repo_id=${HF_USER}/so101_test \\\n --policy.type=act \\\n --output_dir=outputs/train/act_so101_test \\" }, { "id": "ev-lerobot-il-robots-teleop", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/il_robots.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "The v0.6.1 imitation-learning tutorial teleoperates and records datasets with an SO101 leader/follower pair.", "quote": "In this example, we’ll demonstrate how to teleoperate the SO101 robot." }, { "id": "ev-smolvla-paper-so101", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/html/2506.01844v1", "date": "2025-06-02", "verified_on": "2026-10-24", "summary": "SmolVLA paper, Table 4: real-world SO101 Pick-Place-Lego benchmark, ACT vs SmolVLA, in- and out-of-distribution.", "quote": "Similarly, on SO101 (see Table 4), SmolVLA surpasses ACT in both in-distribution and out-of-distribution (OOD) settings." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/act" }, { "robot": "so101", "policy": "diffusion", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "LeRobot's Diffusion Policy defaults target simulation; real SO-101 deployment in #2506 only worked after changing crop_shape (None or matched to image size), vision backbone and inference steps." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-ymatari-diffusion-so101", "ev-hub-jliu-diffusion-so101" ] }, "inference_real": { "state": "works-with-caveats", "evidence": [ "ev-gh-lerobot-2506" ] } }, "evidence": [ { "id": "ev-hub-ymatari-diffusion-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/ymatari/diffusion_so101_cleanup_table_2", "date": "2025-08-27", "verified_on": "2026-10-24", "summary": "config.json type=diffusion; train_config.json dataset.repo_id=ymatari/cleanup-table-2, whose meta/info.json robot_type=so101_follower (20 episodes)." }, { "id": "ev-hub-jliu-diffusion-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/jliu6718/diffusion-so101-place_brick", "date": "2026-01-09", "verified_on": "2026-10-24", "summary": "config.json type=diffusion; train_config.json dataset.repo_id=jliu6718/diffusion-so101-place_brick, whose meta/info.json robot_type=so101_follower (30 episodes)." }, { "id": "ev-gh-lerobot-2506", "tier": "community-reported", "type": "github-issue", "url": "https://github.com/huggingface/lerobot/issues/2506", "date": "2025-11-23", "verified_on": "2026-10-24", "summary": "Issue #2506: a user deployed Diffusion Policy on the SO101 follower after changing the sim-oriented defaults (crop_shape, vision backbone, inference steps); maintainer pkooij confirmed and closed it.", "quote": "I have successfully deployed Diffusion Policy on the SO101 Follower using the 100+ episode dataset, following your guidance." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/diffusion" }, { "robot": "so101", "policy": "evo1", "state": "works", "framework": "evo-1", "capabilities": { "inference_real": { "state": "works", "notes": "Upstream SO100/SO101 LeRobot inference lives on the evo1-lerobot branch of MINT-SJTU/Evo-1.", "evidence": [ "ev-evo1-readme-so100-so101" ] } }, "evidence": [ { "id": "ev-evo1-readme-so100-so101", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/MINT-SJTU/Evo-1/blob/5fd14b015013c4fd0aacf5f8f48f868ca9b870a2/README.md", "date": "2026-08-18", "verified_on": "2026-10-24", "summary": "Official Evo-1 repo README: SO100/SO101 inference via the LeRobot framework (instructions on its evo1-lerobot branch).", "quote": "Added Evo-1 inference in the LeRobot framework for SO100/SO101" } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/evo1" }, { "robot": "so101", "policy": "fastwam", "state": "partial", "partial_scope": "Upstream docs only give a generic real-robot rollout command for so101_follower; no SO-101 training recipe, real-robot checkpoint or result is published. The only SO-101 checkpoint found is trained on simulated SO-101 data.", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "inference_real": { "state": "partial", "notes": "Command only, with a placeholder checkpoint.", "evidence": [ "ev-lerobot-fastwam-doc" ] }, "train": { "state": "partial", "notes": "Simulated SO-101 (MuJoCo) data only.", "evidence": [ "ev-hub-alexhegit-fastwam-so101-sim" ] } }, "evidence": [ { "id": "ev-lerobot-fastwam-doc", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/fastwam.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 FastWAM guide gives a real-robot rollout command for so101_follower with a placeholder checkpoint (your-org/fastwam-real-robot); no SO-101 checkpoint, training recipe or result is published.", "quote": "For real-robot rollout, use the same checkpoint path:\n\n```bash\nlerobot-rollout \\\n --robot.type=so101_follower \\" }, { "id": "ev-hub-alexhegit-fastwam-so101-sim", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/alexhegit/so101-simstudio-lab01-pnp-fastwam", "date": "2026-09-17", "verified_on": "2026-10-24", "summary": "config.json type=fastwam; train_config.json dataset.repo_id=alexhegit/so101-simstudio-lab01-pnp, whose meta/info.json robot_type=so101_mujoco (50 episodes, simulated SO-101)." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/fastwam" }, { "robot": "so101", "policy": "groot-n1-7", "state": "works", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "train": { "state": "works", "evidence": [ "ev-lerobot-groot-doc-train", "ev-isaac-groot-so100-example" ] }, "inference_real": { "state": "works", "evidence": [ "ev-lerobot-groot-doc-rollout", "ev-isaac-groot-so100-example" ] } }, "evidence": [ { "id": "ev-lerobot-groot-doc-train", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/groot.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 GR00T N1.7 guide: the fine-tuning example (--policy.type=groot, base nvidia/GR00T-N1.7-3B) uses the new_embodiment tag for the SO-101.", "quote": "This command is using the `new_embodiment` flag, which is used for the SO-101 robot" }, { "id": "ev-lerobot-groot-doc-rollout", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/groot.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 GR00T N1.7 guide: real-hardware evaluation with lerobot-rollout on so101_follower (optionally with RTC).", "quote": " --policy.base_model_path=nvidia/GR00T-N1.7-3B \\\n --policy.n_action_steps=8 \\\n --robot.type=so101_follower \\" }, { "id": "ev-isaac-groot-so100-example", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/NVIDIA/Isaac-GR00T/blob/n1.7-release/examples/SO100/README.md", "date": "2026-04-18", "verified_on": "2026-10-24", "versions": { "isaac-groot": "n1.7-release" }, "summary": "Isaac-GR00T n1.7-release 'SO100' example fine-tunes GR00T-N1.7-3B (NEW_EMBODIMENT) on izuluaga/finish_sandwich (meta/info.json robot_type=so101_follower) and runs closed-loop eval with --robot.type=so101_follower.", "quote": "This guide shows how to finetune dataset collected from [SO100](https://huggingface.co/docs/lerobot/en/so101) robot, and evaluate the model on the real robot." }, { "id": "ev-isaac-groot-eval-so100", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/NVIDIA/Isaac-GR00T/blob/n1.7-release/gr00t/eval/real_robot/SO100/eval_so100.py", "date": "2026-04-18", "verified_on": "2026-10-24", "versions": { "isaac-groot": "n1.7-release" }, "summary": "Isaac-GR00T n1.7-release real-robot eval script (LeRobot RobotConfig + GR00T policy server) states it targets both SO100 and SO101; it imports LeRobot so100_follower and so101_follower configs.", "quote": "This script runs closed-loop policy evaluation on the SO100 / SO101 robots\nusing the GR00T Policy API." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/groot-n1-7" }, { "robot": "so101", "policy": "lingbot-va", "state": "partial", "partial_scope": "Training on SO-101 data is evidenced (an upstream contributor reproduced post-training on a user's SO-101 dataset); on-robot inference is reported only with low success and jitter in an open, unresolved issue.", "framework": "lingbot-va", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "The reporter's SO-101 deployments (open-loop and closed-loop with KV-cache updates) produced random or back-and-forth motion and only one, apparently accidental, success. A later commenter attributes the jitter to an action-reshaping bug in the reporter's client script; no SO-101 success is confirmed in the thread." ], "capabilities": { "train": { "state": "works", "notes": "Upstream contributor converted the 83-episode SO-101 dataset and trained it with the official post-training code (8 GPUs, reasonable visualizations after ~200 steps).", "evidence": [ "ev-gh-lingbotva-29", "ev-dataset-pick-n-place-sq" ] } }, "evidence": [ { "id": "ev-gh-lingbotva-29", "tier": "community-reported", "type": "github-issue", "url": "https://github.com/Robbyant/lingbot-va/issues/29", "date": "2026-02-26", "verified_on": "2026-10-25", "summary": "Issue 'Finetuning on custom pick and place dataset on SO-101 arm' (open): user fine-tuned robbyant/lingbot-va-base on an SO-101 dataset and deployed it with poor results; contributor lilin-hitcrt reproduced training on the converted dataset and shared a debug config (wan_va/configs/va_demo_train_cfg.py).", "quote": "I've converted your dataset to the proper format and tested it with our training code. Following the provided data format and default parameters, the model trains successfully on 8 GPUs and shows good visualization results after ~200 steps." }, { "id": "ev-dataset-pick-n-place-sq", "tier": "community-reported", "type": "community-post", "url": "https://huggingface.co/datasets/AkashKarnatak/pick-n-place-sq", "verified_on": "2026-10-25", "summary": "The dataset used in the issue: meta/info.json robot_type=so101_follower, LeRobot v3.0 format, 83 episodes, top + wrist cameras." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/lingbot-va" }, { "robot": "so101", "policy": "molmoact2", "state": "works-with-caveats", "framework": "lerobot", "framework_version_tested": "0.6.1", "caveats": [ "The MolmoAct2-SO100_101 weights expect the pre-LeRobot-0.5.0 joint calibration convention: lerobot/MolmoAct2-SO100_101-LeRobot includes the frame correction, but self-converted or fine-tuned checkpoints must set joint_signs/joint_offsets or the arm may move in the wrong direction.", "LeRobot's rollout example is written for so100_follower; SO-101 coverage rests on the official model card's SO-100/101 scope." ], "capabilities": { "inference_real": { "state": "works-with-caveats", "evidence": [ "ev-molmoact2-so100-101-card", "ev-lerobot-molmoact2-calib" ] }, "train": { "state": "works-with-caveats", "evidence": [ "ev-molmoact2-so100-101-card", "ev-lerobot-molmoact2-calib" ] } }, "evidence": [ { "id": "ev-molmoact2-so100-101-card", "tier": "upstream-documented", "type": "model-card", "url": "https://huggingface.co/allenai/MolmoAct2-SO100_101/blob/152569fe57914d97be91055800035f54e250d009/README.md", "date": "2026-05-23", "verified_on": "2026-10-24", "summary": "Ai2's official MolmoAct2-SO100_101 model card: fine-tuned on an SO-100/101 data mixture, intended for SO-100/101 inference and further fine-tuning.", "quote": "This checkpoint is fine-tuned on the SO-100/101 mixture with absolute joint-pose control and annotated language instructions. It is intended for both further fine-tuning and SO-100/101 policy inference." }, { "id": "ev-lerobot-molmoact2-calib", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/molmoact2.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 MolmoAct2 guide warns that the SO-100/101 checkpoint uses the pre-0.5.0 joint calibration convention; the converted LeRobot checkpoint includes the frame correction, self-converted/fine-tuned ones need joint_signs/joint_offsets.", "quote": "The MolmoAct2-SO100_101 checkpoint was trained on data that uses a different\njoint calibration convention than LeRobot >= 0.5.0. Without a frame\ncorrection, the arm may move in the wrong direction." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/molmoact2" }, { "robot": "so101", "policy": "multitask-dit", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-tchuang-dit-so101", "ev-hub-henryzhang-dit-so101" ] } }, "evidence": [ { "id": "ev-hub-tchuang-dit-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/tc-huang/multi_task_dit_mahjong_v1_tray07_twos_dart", "date": "2026-10-13", "verified_on": "2026-10-24", "summary": "config.json type=multi_task_dit; train_config.json dataset.repo_id=tc-huang/mahjong_v1_tray07_twos_dart, whose meta/info.json robot_type=so101_follower (1000 episodes)." }, { "id": "ev-hub-henryzhang-dit-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/HenryZhang/VLAReplica_dit_v4_expert", "date": "2026-04-15", "verified_on": "2026-10-24", "summary": "config.json type=multi_task_dit; train_config.json dataset.repo_id=HenryZhang/VLAReplicaMerge_v4, whose meta/info.json robot_type=so101_follower (250 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/multitask-dit" }, { "robot": "so101", "policy": "pi0-fast", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-ryota-pi0fast-so101", "ev-hub-observabot-pi0fast-so101" ] } }, "evidence": [ { "id": "ev-hub-ryota-pi0fast-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/RyotaYoneyamaNN/pi0fast_so101_rabbit2bear_lab", "date": "2025-08-25", "verified_on": "2026-10-24", "summary": "config.json type=pi0fast; train_config.json dataset.repo_id=RyotaYoneyamaNN/move_rabbit2bear_lab2, whose meta/info.json robot_type=so101_follower (100 episodes). Uses the earlier 'pi0fast' policy type (pre-'pi0_fast' LeRobot implementation)." }, { "id": "ev-hub-observabot-pi0fast-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/observabot/pi0fast_so101_die_mat4_b24_lr1e-2_cs100_nas100_robo", "verified_on": "2026-10-24", "summary": "config.json type=pi0fast; train_config.json dataset.repo_id=observabot/so101_die_mat4, whose meta/info.json robot_type=so101_follower (217 episodes). Uses the earlier 'pi0fast' policy type." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/pi0-fast" }, { "robot": "so101", "policy": "pi0", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-amindt-pi0-so101", "ev-hub-vikrant-pi0-so101", "ev-hub-shani-pi0-so101" ] } }, "evidence": [ { "id": "ev-hub-amindt-pi0-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/amindt/pi0_so101_pickandplace_redblue", "date": "2025-11-24", "verified_on": "2026-10-24", "summary": "config.json type=pi0; train_config.json dataset.repo_id=amindt/so101_pickandplace_redblue, whose meta/info.json robot_type=so101_follower (50 episodes)." }, { "id": "ev-hub-vikrant-pi0-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/vikrant1998/pi0_so101_v3", "date": "2025-12-11", "verified_on": "2026-10-24", "summary": "config.json type=pi0; train_config.json dataset.repo_id=vikrant1998/so101_v3_dataset, whose meta/info.json robot_type=so101_follower (50 episodes)." }, { "id": "ev-hub-shani-pi0-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Shani123/pi0_so101_pickup_toothpicks_2_plus_recovery_20000", "date": "2025-09-25", "verified_on": "2026-10-24", "summary": "config.json type=pi0; train_config.json dataset.repo_id=Shani123/pickup_toothpicks_2_plus_recovery, whose meta/info.json robot_type=so101_follower (66 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/pi0" }, { "robot": "so101", "policy": "pi05", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "The quantified report (about 90% on one pick-and-place task) did not record the number of trials and ran through VLASH's PI05Policy on lerobot 0.4.1, not stock lerobot-rollout.", "The second report (pi05_base action-expert-only fine-tune, 50 episodes) shows only qualitative live-rollout videos and states the policy is not yet robust across the full workspace.", "Both are single-arm SO-101 setups with task-specific fine-tuning and fixed camera placement." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-00ri-pi05-so101", "ev-hub-tinkhireeva-pi05-so101", "ev-hub-zz4321-pi05-so101", "ev-card-dongyoonkim-pi05-so101", "ev-card-zkxuerb-pi05-so101" ] }, "inference_real": { "state": "works-with-caveats", "notes": "About 90% observed success (trial count not recorded) on one task; a second checkpoint has phone-recorded live-evaluation videos with remote A100 inference (~1.3 s round-trip per 50-action chunk).", "evidence": [ "ev-card-dongyoonkim-pi05-so101", "ev-card-zkxuerb-pi05-so101" ] } }, "evidence": [ { "id": "ev-hub-00ri-pi05-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/00ri/so101_megamix_v1_pi05", "date": "2026-01-04", "verified_on": "2026-10-24", "summary": "config.json type=pi05; train_config.json dataset.repo_id=whosricky/so101-megamix-v1, whose meta/info.json robot_type=so101_follower (400 episodes)." }, { "id": "ev-hub-tinkhireeva-pi05-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/tinkhireeva/pi05_so101_pick_place_yellow_objects_10_locations", "date": "2025-10-31", "verified_on": "2026-10-24", "summary": "config.json type=pi05; train_config.json dataset.repo_id=tinkhireeva/so101_pick_place_yellow_objects_10_locations, whose meta/info.json robot_type=so101_follower (150 episodes)." }, { "id": "ev-hub-zz4321-pi05-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/zz4321/so101_pi05", "date": "2025-10-12", "verified_on": "2026-10-24", "summary": "config.json type=pi05; train_config.json dataset.repo_id=zz4321/so101_cube, whose meta/info.json robot_type=so101_follower (50 episodes)." }, { "id": "ev-card-dongyoonkim-pi05-so101", "tier": "community-reported", "type": "model-card", "url": "https://huggingface.co/dongyoonkim/so101-pi05-base/blob/56325bafcca9ec3155b8bfdf655be7df17e7f09f/README.md", "date": "2026-08-04", "verified_on": "2026-10-25", "versions": { "lerobot": "0.4.1", "vlash": "22cbabfee0f57874987c75a35a7dac129e695db0", "torch": "2.7.1+cu126" }, "summary": "Card for a pi0.5 checkpoint fully fine-tuned from lerobot/pi05_base on SO-101/SO-100 data reports a closed-loop evaluation on a physical SO-101 follower (2 cameras, step 40000, run via VLASH on an RTX 3090 Ti): about 90% observed success on one task, number of trials not recorded.", "quote": "Closed-loop evaluation on physical hardware, 2026-06-26. This is a qualitative observation on one\ntask and one setup, not a benchmark." }, { "id": "ev-card-zkxuerb-pi05-so101", "tier": "community-reported", "type": "model-card", "url": "https://huggingface.co/zkxuerb/pi05_so101_pick_cube_v1_expert_only/blob/efe2df1b047fa127f8fb58fbb6a221ac0d46236f/README.md", "date": "2026-06-02", "verified_on": "2026-10-25", "summary": "Card for a pi0.5 action-expert-only fine-tune of lerobot/pi05_base (train_config.json dataset zkxuerb/so101_right_arm_pick_cube_v1, meta/info.json robot_type=so_follower, 50 episodes) embeds two live-evaluation videos on a physical SO-ARM101 follower; qualitative only, not robust across the full workspace.", "quote": "Both clips are real-time, recorded on a phone during live evaluation on the\nSO-ARM101 right follower. Inference was served from a Modal A100 endpoint\nwith ~1.3 s round-trip per 50-action chunk." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/pi05" }, { "robot": "so101", "policy": "smolvla", "state": "works", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "train": { "state": "works", "evidence": [ "ev-smolvla-paper-so101" ] }, "inference_real": { "state": "works", "evidence": [ "ev-lerobot-smolvla-doc", "ev-smolvla-paper-so101" ] }, "teleop_record": { "state": "works", "evidence": [ "ev-smolvla-paper-datasets" ] } }, "evidence": [ { "id": "ev-lerobot-smolvla-doc", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/smolvla.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 SmolVLA guide: fine-tune smolvla_base, then run the fine-tuned model on the real robot with lerobot-rollout --robot.type=so101_follower; its figure shows real-world SO101 results.", "quote": "lerobot-rollout \\\n --strategy.type=base \\\n --robot.type=so101_follower \\" }, { "id": "ev-smolvla-paper-so101", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/html/2506.01844v1", "date": "2025-06-02", "verified_on": "2026-10-24", "summary": "SmolVLA paper, Table 4: real-world SO101 Pick-Place-Lego benchmark, ACT vs SmolVLA, in- and out-of-distribution.", "quote": "Similarly, on SO101 (see Table 4), SmolVLA surpasses ACT in both in-distribution and out-of-distribution (OOD) settings." }, { "id": "ev-smolvla-paper-datasets", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/html/2506.01844v1", "date": "2025-06-02", "verified_on": "2026-10-24", "summary": "SmolVLA paper: real-world evaluation datasets were recorded on SO-100 (three tasks) and SO-101 (one task).", "quote": "For real-world evaluation, we collected three datasets using the SO-100 robot arm" } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/smolvla" }, { "robot": "so101", "policy": "vla-jepa", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-koenvanwijk-vlajepa-so101", "ev-hub-binhpham-vlajepa-so101" ] } }, "evidence": [ { "id": "ev-hub-koenvanwijk-vlajepa-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/koenvanwijk/vla-jepa-so101-finetune", "date": "2026-06-07", "verified_on": "2026-10-24", "summary": "config.json type=vla_jepa; train_config.json dataset.repo_id=koenvanwijk/find-coin50-1, whose meta/info.json robot_type=so101_follower (50 episodes)." }, { "id": "ev-hub-binhpham-vlajepa-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/binhpham/naive-bench-vla-jepa", "date": "2026-07-01", "verified_on": "2026-10-24", "summary": "config.json type=vla_jepa; train_config.json dataset.repo_id=binhpham/naive-bench, whose meta/info.json robot_type=so101_follower (312 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/vla-jepa" }, { "robot": "so101", "policy": "vqbet", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-ljonas-vqbet-so101" ] } }, "evidence": [ { "id": "ev-hub-ljonas-vqbet-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/ljonas/red-cube-cup-vqbet", "date": "2026-01-08", "verified_on": "2026-10-24", "summary": "config.json type=vqbet; train_config.json dataset.repo_id=ljonas/red-cube-cup, whose meta/info.json robot_type=so101_follower (50 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/vqbet" }, { "robot": "so101", "policy": "wall-oss", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-aractingi-wallx-so101", "ev-hub-pepijn-wallx-so101" ] } }, "evidence": [ { "id": "ev-hub-aractingi-wallx-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/aractingi/wall_x_3", "date": "2025-12-15", "verified_on": "2026-10-24", "summary": "config.json type=wall_x; train_config.json dataset.repo_id=imstevenpmwork/thanos_picking_power_gem_1749731584242992, whose meta/info.json robot_type=so101_follower (51 episodes)." }, { "id": "ev-hub-pepijn-wallx-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/pepijn223/wall_x", "date": "2025-12-09", "verified_on": "2026-10-24", "summary": "config.json type=wall_x; train_config.json dataset.repo_id=imstevenpmwork/thanos_picking_power_gem_1749731584242992, whose meta/info.json robot_type=so101_follower (51 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/wall-oss" }, { "robot": "so101", "policy": "xvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-whosricky-xvla-so101", "ev-hub-zhiqiang-xvla-so101" ] } }, "evidence": [ { "id": "ev-hub-whosricky-xvla-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/whosricky/xvla-so101-pick-red-cube-2cam", "date": "2025-12-12", "verified_on": "2026-10-24", "summary": "config.json type=xvla; train_config.json dataset.repo_id=whosricky/pick-red-cube-v3, whose meta/info.json robot_type=so101_follower (182 episodes)." }, { "id": "ev-hub-zhiqiang-xvla-so101", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/zhiqiang3/xvla_so101_pickplace_front_wrist_camera_merge_all", "date": "2025-12-15", "verified_on": "2026-10-24", "summary": "config.json type=xvla; train_config.json dataset.repo_id=zhiqiang3/so101_pickplace_front_wrist_camera_merge_all, whose meta/info.json robot_type=so101_follower (170 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/so101/xvla" }, { "robot": "trossen-widowx", "policy": "act", "state": "works", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-trossen-docs-train-eval", "ev-trossen-act-tshirt-card" ] }, "inference_real": { "state": "works", "evidence": [ "ev-trossen-docs-train-eval", "ev-trossen-plugin-readme", "ev-trossen-act-tshirt-card" ] } }, "evidence": [ { "id": "ev-trossen-docs-train-eval", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://docs.trossenrobotics.com/trossen_arm/v1.9/tutorials/lerobot_plugin/train_and_evaluate.html", "verified_on": "2026-10-24", "summary": "Trossen's LeRobot-plugin tutorial trains ACT (--policy.type=act) on Trossen AI Solo/Stationary/Mobile (WidowX AI) datasets and evaluates it on the robot with lerobot-record --robot.type=widowxai_follower_robot / bi_widowxai_follower_robot.", "quote": "We specified the policy with --policy.type=act, which loads configurations from configuration_act.py." }, { "id": "ev-trossen-plugin-readme", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/TrossenRobotics/lerobot_trossen/blob/337ed74df1ee617ddfb5987a9f7464cf3089c4a2/README.md", "date": "2026-09-16", "verified_on": "2026-10-24", "summary": "Official Trossen plugin README (requires lerobot >= 0.6.0) shows an ACT policy evaluated on a single WidowX AI via lerobot-record --policy.path.", "quote": "Evaluate a trained policy by recording 2 episodes of a cube pickup task with a single WidowX AI robot using the OpenCV camera interface." }, { "id": "ev-trossen-act-tshirt-card", "tier": "upstream-documented", "type": "model-card", "url": "https://huggingface.co/TrossenRoboticsCommunity/act-tshirt-folding-lerobot/blob/a71a4013b54206b0ca15fc486a69ecd33f5ee7f3/README.md", "date": "2026-07-01", "verified_on": "2026-10-24", "summary": "Trossen-published ACT checkpoint (config.json type=act; train_config dataset hamza-trossen/tshirt-folding-mobile-stationary, whose meta/info.json robot_type=bi_widowxai_follower_robot) with an on-robot eval video and a 6/10 success rate on bimanual t-shirt folding.", "quote": "**ACT** (Action Chunking Transformer) policy for bimanual t-shirt folding on the Trossen AI **stationary** (WidowX AI) platform. Trained with **lerobot**." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/trossen-widowx/act" }, { "robot": "trossen-widowx", "policy": "diffusion", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "The single report concerns a policy deliberately overfit to one pick-and-place trajectory, not a generalizing policy." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-lerobot-issue-1811" ] }, "inference_real": { "state": "works", "evidence": [ "ev-lerobot-issue-1811" ] } }, "evidence": [ { "id": "ev-lerobot-issue-1811", "tier": "community-reported", "type": "github-issue", "url": "https://github.com/huggingface/lerobot/issues/1811", "date": "2025-08-29", "verified_on": "2026-10-24", "summary": "A user fine-tuning on a Trossen WidowX AI pick-and-place dataset reports that ACT and Diffusion Policy overfit to one trajectory and work when rolled out on the physical robot (the issue itself is about π0-FAST failing to converge).", "quote": "I was able to get ACT and diffusion policy to overfit on the same dataset and when rolled out on the physical robot works perfectly unlike pi0Fast" } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/trossen-widowx/diffusion" }, { "robot": "trossen-widowx", "policy": "groot-n1-7", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "framework_version_tested": "0.6.0", "capabilities": { "train": { "state": "works", "evidence": [ "ev-trossen-groot-ckpt" ] } }, "evidence": [ { "id": "ev-trossen-groot-ckpt", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/TrossenRobotics/groot-n17-block-transfer", "date": "2026-07-17", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.0" }, "summary": "Trossen-org checkpoint: config.json type=groot, base_model_path nvidia/GR00T-N1.7-3B; train_config dataset TrossenRoboticsCommunity/stationary-block-transfer-lerobot-v3, whose meta/info.json robot_type=bi_widowxai_follower_robot (card: 4 cameras, LeRobot version 0.6.0). Auto-generated card, no on-robot results." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/trossen-widowx/groot-n1-7" }, { "robot": "trossen-widowx", "policy": "pi0", "state": "works-with-caveats", "framework": "openpi", "caveats": [ "Trossen labels its openpi tutorial experimental, with minimal testing on the Trossen AI setup.", "Fine-tuning is required (the π0 base mixture has no Trossen AI data); in the vendor's tests the 50-episode fine-tune failed on a different block size/colour, a wooden block, and under disturbances.", "Uses Trossen's openpi fork, which trains on LeRobot Dataset v2 (LeRobot v0.1.0); v3.0 datasets are not compatible with that training pipeline." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-trossen-openpi-tutorial" ] }, "inference_real": { "state": "works", "evidence": [ "ev-trossen-openpi-tutorial" ] } }, "evidence": [ { "id": "ev-trossen-openpi-tutorial", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://docs.trossenrobotics.com/trossen_arm/v1.9/tutorials/openpi.html", "verified_on": "2026-10-24", "summary": "Trossen's openpi tutorial fine-tunes π0 (LoRA) on a 50-episode bimanual WidowX AI handover dataset, serves it with openpi's policy server and drives the arms through the LeRobot plugin client; reports one success and three failure scenarios.", "quote": "openpi_trossen_ai_red_block [success] : Robot successfully picks up and transfers red block in second try." }, { "id": "ev-trossen-pi0-openpi-ckpt", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/TrossenRoboticsCommunity/pi0-block-transfer-trossen-ai-openpi", "verified_on": "2026-10-24", "summary": "Trossen-published openpi (orbax params + assets/trossen/norm_stats.json) π0 checkpoint for Trossen AI block transfer, linked from the tutorial as the fine-tuned π0 checkpoint." }, { "id": "ev-trossen-pi0-lerobot-ckpt", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/TrossenRoboticsCommunity/pi0-handover-cube-widowxai", "date": "2025-08-05", "verified_on": "2026-10-24", "summary": "LeRobot route (training only): config.json type=pi0; train_config dataset TrossenRoboticsCommunity/bimanual-widowxai-handover-cube, whose meta/info.json robot_type=bi_widowxai_follower. Card is auto-generated; no on-robot results." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/trossen-widowx/pi0" }, { "robot": "trossen-widowx", "policy": "pi05", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Trossen recommends LeRobot async inference (policy server + robot client) for π0.5; synchronous lerobot-record runs the policy in the control loop, which large VLAs are often too slow for.", "Needs the lerobot[pi] extra (transformers/peft), which the plugin's base install omits.", "Vendor on-robot results are qualitative so far (no published success-rate counts)." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-trossen-pi05-lerobot-card", "ev-trossen-pi05-pour-card" ] }, "inference_real": { "state": "works", "evidence": [ "ev-trossen-plugin-readme", "ev-trossen-pi05-lerobot-card", "ev-trossen-pi05-pour-card" ] } }, "evidence": [ { "id": "ev-trossen-plugin-readme", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/TrossenRobotics/lerobot_trossen/blob/337ed74df1ee617ddfb5987a9f7464cf3089c4a2/README.md", "date": "2026-09-16", "verified_on": "2026-10-24", "summary": "Official Trossen plugin README runs a π0.5 checkpoint (--policy_type=pi05, TrossenRoboticsCommunity pi05-block-transfer-lerobot) on a bimanual WidowX AI (bi_widowxai_follower_robot) via LeRobot async inference, lerobot[async,pi]>=0.6.0.", "quote": "This is the recommended path for slow VLA policies (π₀, π₀.₅, SmolVLA): the client keeps the robot control loop responsive while the server runs inference, and overlapping action chunks are blended on the client (real-time chunking)." }, { "id": "ev-trossen-pi05-lerobot-card", "tier": "upstream-documented", "type": "model-card", "url": "https://huggingface.co/TrossenRoboticsCommunity/pi05-block-transfer-lerobot/blob/ef188f51fd55da409f315472fe1e91e457bb6162/README.md", "date": "2026-06-03", "verified_on": "2026-10-24", "summary": "Trossen-published π0.5 LoRA fine-tune (config.json type=pi05, base lerobot/pi05_base) trained with LeRobot on bimanual WidowX AI block-transfer data, with an on-robot async-eval video; card says 'policy confirmed working on hardware', rep count TODO.", "quote": "π₀.₅ fine-tune for bimanual red-cube handover / block transfer on the Trossen AI" }, { "id": "ev-trossen-pi05-pour-card", "tier": "upstream-documented", "type": "model-card", "url": "https://huggingface.co/TrossenRoboticsCommunity/bimanual-widowxai-pour-items-red-cup-pi05/blob/e4281ebd9506852c8f9526ba3553181e4651a853/README.md", "date": "2026-09-25", "verified_on": "2026-10-24", "summary": "Trossen-published π0.5 checkpoint (config.json type=pi05, LeRobot/PyTorch, 0.5.1-era processor naming) fine-tuned on 100 bimanual WidowX AI pouring episodes; evaluated qualitatively on the real rig.", "quote": "Evaluated on the real robot — the bimanual WidowX AI stationary rig — and reported by the operator as performing the pouring task well." }, { "id": "ev-trossen-openpi-tutorial", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://docs.trossenrobotics.com/trossen_arm/v1.9/tutorials/openpi.html", "verified_on": "2026-10-24", "summary": "Separate route: Trossen's experimental openpi tutorial (Trossen fork of openpi) also fine-tunes π0.5 on bimanual WidowX AI data and reports a successful red-block transfer.", "quote": "openpi_trossen_ai_pi05_red_block [success] : π0.5 model successfully picks up and transfers red block on first try." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/trossen-widowx/pi05" }, { "robot": "trossen-widowx", "policy": "smolvla", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "Trossen documents the SmolVLA evaluation path but has published no on-robot SmolVLA results; its SmolVLA checkpoints carry auto-generated cards only.", "Needs the lerobot[smolvla] extra, which the plugin's base install omits; Trossen recommends async inference for VLA policies." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-trossen-smolvla-ckpt" ] }, "inference_real": { "state": "works", "evidence": [ "ev-trossen-plugin-readme", "ev-trossen-docs-async" ] } }, "evidence": [ { "id": "ev-trossen-plugin-readme", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/TrossenRobotics/lerobot_trossen/blob/337ed74df1ee617ddfb5987a9f7464cf3089c4a2/README.md", "date": "2026-09-16", "verified_on": "2026-10-24", "summary": "Official Trossen plugin README names SmolVLA among the VLA policies that run on WidowX AI robots via lerobot-record or async inference, given the extra dependencies.", "quote": "**VLA policies (π₀, π₀.₅, SmolVLA) need extra dependencies** (transformers/peft) that the base install omits" }, { "id": "ev-trossen-docs-async", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://docs.trossenrobotics.com/trossen_arm/main/tutorials/lerobot_plugin/async_inference.html", "verified_on": "2026-10-24", "summary": "Trossen async-inference tutorial for Trossen AI kits covers serving SmolVLA (swap the pi extra for smolvla).", "quote": "smolvla: required for SmolVLA policies." }, { "id": "ev-trossen-smolvla-ckpt", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/TrossenRoboticsCommunity/smolvla-handover-cube-widowxai", "date": "2025-07-31", "verified_on": "2026-10-24", "summary": "config.json type=smolvla (base lerobot/smolvla_base); train_config dataset TrossenRoboticsCommunity/bimanual-widowxai-handover-cube_01, whose meta/info.json robot_type=bi_widowxai_follower. Trossen also publishes SmolVLA checkpoints on trossen_ai_stationary datasets." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/trossen-widowx/smolvla" }, { "robot": "trossen-widowx", "policy": "vla-jepa", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-luisd-vlajepa" ] } }, "evidence": [ { "id": "ev-hub-luisd-vlajepa", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/luisd7271/vla-jepa-widgetAssembly", "date": "2026-07-10", "verified_on": "2026-10-25", "summary": "config.json type=vla_jepa (init lerobot/VLA-JEPA-Pretrain); train_config.json dataset luisd7271/stationary-WidgetAssembly-Lerobot-act-v3 (10000 steps), whose meta/info.json robot_type=bi_widowxai_follower_robot (25 episodes, Trossen stationary bimanual WidowX AI)." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/trossen-widowx/vla-jepa" }, { "robot": "trossen-widowx", "policy": "xvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-qownscks-xvla", "ev-hub-nyamamoto-xvla" ] } }, "evidence": [ { "id": "ev-hub-qownscks-xvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/qownscks/xvlwidow", "date": "2026-03-10", "verified_on": "2026-10-25", "summary": "config.json type=xvla (init lerobot/xvla-widowx); train_config.json dataset qownscks/eggplant (60000 steps), whose meta/info.json robot_type=widowxai_follower_robot (15 episodes)." }, { "id": "ev-hub-nyamamoto-xvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/nyamamoto/20260226-daiv-red-blue-cube-xvla", "date": "2026-02-26", "verified_on": "2026-10-25", "summary": "config.json type=xvla (init lerobot/xvla-widowx); train_config.json dataset nyamamoto/20260218-red-blue-cube (100000 steps), whose meta/info.json robot_type=widowxai_follower_robot (61 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/trossen-widowx/xvla" }, { "robot": "unitree-g1", "policy": "act", "state": "works-with-caveats", "caveats": [ "Documented by Unitree in its unitree_IL_lerobot repository, which pins LeRobot as a git submodule (commit a5b29d4, November 2025) and deploys with its own eval_robot/eval_g1.py script, not through LeRobot's native unitree_g1 robot or lerobot-rollout.", "The documented workflow covers the G1's arms and dexterous hands (--arm G1_29/G1_23, --ee dex1/dex3/inspire1/brainco); it does not document leg or locomotion control." ], "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-unitree-il-train" ] }, "inference_real": { "state": "works", "evidence": [ "ev-unitree-il-eval" ] } }, "evidence": [ { "id": "ev-unitree-il-train", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/unitreerobotics/unitree_IL_lerobot/blob/41c2805742de879ddab2d8d6beaeaf215f876395/README.md", "verified_on": "2026-10-24", "versions": { "lerobot": "a5b29d4" }, "summary": "Unitree's README gives a lerobot_train.py command for ACT (--policy.type=act) on its G1 Dex3 dataset.", "quote": "--dataset.repo_id=unitreerobotics/G1_Dex3_ToastedBread_Dataset \\\n --policy.push_to_hub=false \\\n --policy.type=act" }, { "id": "ev-unitree-il-eval", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/unitreerobotics/unitree_IL_lerobot/blob/41c2805742de879ddab2d8d6beaeaf215f876395/README.md", "verified_on": "2026-10-24", "versions": { "lerobot": "a5b29d4" }, "summary": "Unitree documents real-robot testing of a trained LeRobot policy on the G1 via eval_g1.py (its example checkpoint path is a diffusion run).", "quote": "To test your trained model on a real robot, you can use the eval_g1.py script located in the eval_robot folder." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/unitree-g1/act" }, { "robot": "unitree-g1", "policy": "diffusion", "state": "works-with-caveats", "caveats": [ "Documented by Unitree in its unitree_IL_lerobot repository, which pins LeRobot as a git submodule (commit a5b29d4, November 2025) and deploys with its own eval_robot/eval_g1.py script, not through LeRobot's native unitree_g1 robot or lerobot-rollout.", "The documented workflow covers the G1's arms and dexterous hands (--arm G1_29/G1_23, --ee dex1/dex3/inspire1/brainco); it does not document leg or locomotion control." ], "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-unitree-il-train" ] }, "inference_real": { "state": "works", "evidence": [ "ev-unitree-il-eval" ] } }, "evidence": [ { "id": "ev-unitree-il-train", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/unitreerobotics/unitree_IL_lerobot/blob/41c2805742de879ddab2d8d6beaeaf215f876395/README.md", "verified_on": "2026-10-24", "versions": { "lerobot": "a5b29d4" }, "summary": "Unitree's README gives a lerobot_train.py command for Diffusion Policy (--policy.type=diffusion) on its G1 Dex3 dataset.", "quote": "--dataset.repo_id=unitreerobotics/G1_Dex3_ToastedBread_Dataset \\\n --policy.push_to_hub=false \\\n --policy.type=diffusion" }, { "id": "ev-unitree-il-eval", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/unitreerobotics/unitree_IL_lerobot/blob/41c2805742de879ddab2d8d6beaeaf215f876395/README.md", "verified_on": "2026-10-24", "versions": { "lerobot": "a5b29d4" }, "summary": "Unitree documents real-robot testing of a trained LeRobot policy on the G1 via eval_g1.py (its example checkpoint path is a diffusion run).", "quote": "To test your trained model on a real robot, you can use the eval_g1.py script located in the eval_robot folder." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/unitree-g1/diffusion" }, { "robot": "unitree-g1", "policy": "groot-n1-7", "state": "works-with-caveats", "caveats": [ "Documented in NVIDIA's Isaac-GR00T repository (embodiment tag UNITREE_G1_SONIC), not through LeRobot: whole-body motion is executed by the GR00T Whole-Body Control / GEAR-SONIC robot-side controller, with GR00T N1.7 served from the GR00T policy server.", "For the pickup task NVIDIA used an in-house UMI gripper design instead of a stock G1 hand, and notes that similar tasks may need task-appropriate grippers.", "The example landed on Isaac-GR00T main after the n1.7-release tag; it is not part of a tagged release." ], "framework": "isaac-groot", "capabilities": { "train": { "state": "works", "notes": "Fine-tuned from nvidia/GR00T-N1.7-3B with examples/finetune.sh on G1 demonstrations collected with SONIC teleoperation (LeRobot v2 data format).", "evidence": [ "ev-gr00t-wbc-g1", "ev-gr00t-wbc-workflow" ] }, "inference_real": { "state": "works", "notes": "Closed-loop real-robot evaluation: 159/216 (73.6%) on walk-to-table-and-pick-up across 8 objects, 3 table heights and 9 placements; 8/12 (66.7%) on a soda-can-to-trash-can task. Autonomous retries were counted as success.", "evidence": [ "ev-gr00t-wbc-g1", "ev-gr00t-wbc-workflow" ] }, "teleop_record": { "state": "works", "evidence": [ "ev-gr00t-wbc-workflow" ] } }, "evidence": [ { "id": "ev-gr00t-wbc-g1", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/NVIDIA/Isaac-GR00T/blob/1a1837f20538b7d7e21f977a11a5aee14f99803c/examples/GR00TWholeBodyControl/README.md", "date": "2026-07-07", "verified_on": "2026-10-24", "summary": "NVIDIA reports a real-world Unitree G1 mobile-manipulation evaluation of GR00T N1.7 combined with GR00T Whole-Body Control / GEAR-SONIC.", "quote": "This example reports a real-world Unitree G1 evaluation using **Isaac GR00T N1.7** together with **GR00T Whole-Body Control / GEAR-SONIC**." }, { "id": "ev-gr00t-wbc-workflow", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/NVIDIA/Isaac-GR00T/blob/1a1837f20538b7d7e21f977a11a5aee14f99803c/examples/GR00TWholeBodyControl/README.md", "date": "2026-07-07", "verified_on": "2026-10-24", "summary": "The documented data-train-eval workflow: SONIC teleoperation for data, GR00T N1.7 fine-tuning with the UNITREE_G1_SONIC tag, closed-loop evaluation on the real robot.", "quote": "1. Collect G1 demonstrations with SONIC teleoperation.\n2. Fine-tune GR00T N1.7 with `UNITREE_G1_SONIC`.\n3. Run the GR00T policy server and the SONIC robot-side controller.\n4. Evaluate in closed loop on the real robot with video recording." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/unitree-g1/groot-n1-7" }, { "robot": "unitree-g1", "policy": "pi0-fast", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-eunjuri-pi0fast-g1" ] } }, "evidence": [ { "id": "ev-hub-eunjuri-pi0fast-g1", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/eunjuri/pi0-fast_policy_pick_and_place_for_train_no_tac", "date": "2026-02-23", "verified_on": "2026-10-25", "summary": "config.json type=pi0_fast (init lerobot/pi0fast-libero); train_config.json dataset eunjuri/pick_and_place_for_train_no_tac (10000 steps), whose meta/info.json robot_type=Unitree_G1_Inspire (1035 episodes, G1 arm + Inspire hand joints). Dataset card does not say whether it is real or simulated; auto-generated model card." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/unitree-g1/pi0-fast" }, { "robot": "unitree-g1", "policy": "pi0", "state": "works-with-caveats", "caveats": [ "Documented by Unitree in its unitree_IL_lerobot repository, which pins LeRobot as a git submodule (commit a5b29d4, November 2025) and deploys with its own eval_robot/eval_g1.py script, not through LeRobot's native unitree_g1 robot or lerobot-rollout.", "The documented workflow covers the G1's arms and dexterous hands (--arm G1_29/G1_23, --ee dex1/dex3/inspire1/brainco); it does not document leg or locomotion control." ], "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-unitree-il-train" ] }, "inference_real": { "state": "works", "evidence": [ "ev-unitree-il-eval" ] } }, "evidence": [ { "id": "ev-unitree-il-train", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/unitreerobotics/unitree_IL_lerobot/blob/41c2805742de879ddab2d8d6beaeaf215f876395/README.md", "verified_on": "2026-10-24", "versions": { "lerobot": "a5b29d4" }, "summary": "Unitree's README gives a lerobot_train.py command for π0 (--policy.type=pi0) on its G1 Dex3 dataset.", "quote": "--dataset.repo_id=unitreerobotics/G1_Dex3_ToastedBread_Dataset \\\n --policy.push_to_hub=false \\\n --policy.type=pi0" }, { "id": "ev-unitree-il-eval", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/unitreerobotics/unitree_IL_lerobot/blob/41c2805742de879ddab2d8d6beaeaf215f876395/README.md", "verified_on": "2026-10-24", "versions": { "lerobot": "a5b29d4" }, "summary": "Unitree documents real-robot testing of a trained LeRobot policy on the G1 via eval_g1.py (its example checkpoint path is a diffusion run).", "quote": "To test your trained model on a real robot, you can use the eval_g1.py script located in the eval_robot folder." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/unitree-g1/pi0" }, { "robot": "unitree-g1", "policy": "pi05", "state": "works", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "train": { "state": "works", "notes": "Upstream fine-tunes from lerobot/pi05_base on a G1 dataset recorded with the Homunculus exoskeleton (bfloat16, gradient checkpointing, --policy.device=cuda).", "evidence": [ "ev-lerobot-g1-train", "ev-unitree-il-pi05", "ev-hub-xiaopengwu-pi05-g1" ] }, "inference_real": { "state": "works", "notes": "Deployed with lerobot-rollout and --inference.type=rtc on --robot.is_simulation=false. Walking is handled by a separate locomotion controller chosen with --robot.controller (HolosomaLocomotionController in the documented command; GrootLocomotionController is also supported); the page does not present π0.5 as the walking controller.", "evidence": [ "ev-lerobot-g1-rollout" ] }, "teleop_record": { "state": "works", "notes": "The documented recording command runs in MuJoCo (--robot.is_simulation=true) with the Homunculus exoskeleton or a joystick.", "evidence": [ "ev-lerobot-g1-intro" ] } }, "evidence": [ { "id": "ev-lerobot-g1-intro", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/unitree_g1.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot supports the Unitree G1 (29 and 23 DoF) for teleoperation, locomanipulation policy training and simulation.", "quote": "The Unitree G1 humanoid is now supported in LeRobot! You can teleoperate, train locomanipulation policies, test in sim, and more. Both 29 and 23 DoF variants are supported." }, { "id": "ev-lerobot-g1-train", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/unitree_g1.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "The G1 page's 'Training & Inference' section fine-tunes π0.5 (--policy.type=pi05, pretrained lerobot/pi05_base).", "quote": "--policy.type=pi05 \\\n --output_dir=./outputs/pi05_training" }, { "id": "ev-lerobot-g1-rollout", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/unitree_g1.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "Upstream deploys the trained policy on the physical G1 with lerobot-rollout, RTC inference and the Holosoma locomotion controller.", "quote": "--robot.type=unitree_g1 \\\n --robot.is_simulation=false \\\n --robot.controller=HolosomaLocomotionController" }, { "id": "ev-unitree-il-pi05", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/unitreerobotics/unitree_IL_lerobot/blob/41c2805742de879ddab2d8d6beaeaf215f876395/README.md", "verified_on": "2026-10-24", "versions": { "lerobot": "a5b29d4" }, "summary": "Unitree's own LeRobot-based repo documents π0.5 training on its G1 Dex3 dataset (repo pins LeRobot as a submodule at commit a5b29d4).", "quote": "--dataset.repo_id=unitreerobotics/G1_Dex3_ToastedBread_Dataset \\\n --policy.type=pi05" }, { "id": "ev-hub-xiaopengwu-pi05-g1", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/xiaopeng-wu/pi05_unitree_g1", "verified_on": "2026-10-24", "summary": "config.json type=pi05; train_config.json dataset nepyope/unitree_box_move_blue_full (the example dataset linked from the upstream G1 page) whose meta/info.json robot_type=unitree_g1 (550 episodes); pretrained from lerobot/pi05_base, 3000 steps." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/unitree-g1/pi05" }, { "robot": "unitree-g1", "policy": "smolvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-qualia-smolvla-g1", "ev-hub-eunjuri-smolvla-g1" ] } }, "evidence": [ { "id": "ev-hub-qualia-smolvla-g1", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/qualiaadmin/68bf0913-c0a0-4b98-b1d1-c14dc1c55a44", "date": "2026-01-15", "verified_on": "2026-10-25", "summary": "config.json type=smolvla; train_config.json dataset unitreerobotics/G1_Dex1_MountCamera_Dataset (Unitree's own G1 teleoperation dataset), whose meta/info.json robot_type=Unitree_G1 (351 episodes; G1 arm joints kLeftShoulderPitch... plus Dex1 grippers; four cameras). Unitree's robot_type string, not LeRobot's unitree_g1." }, { "id": "ev-hub-eunjuri-smolvla-g1", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/eunjuri/smolvla_red_cup_depth", "date": "2026-05-01", "verified_on": "2026-10-25", "summary": "config.json type=smolvla; train_config.json dataset eunjuri/red_cup_img_depth (10000 steps), whose meta/info.json robot_type=Unitree_G1_Inspire (83 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/unitree-g1/smolvla" }, { "robot": "unitree-g1", "policy": "xvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-eunjuri-xvla-g1" ] } }, "evidence": [ { "id": "ev-hub-eunjuri-xvla-g1", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/eunjuri/xvla_pick_and_place_for_train_no_tac", "date": "2026-02-23", "verified_on": "2026-10-25", "summary": "config.json type=xvla (init lerobot/xvla-base); train_config.json dataset eunjuri/pick_and_place_for_train_no_tac (20000 steps), whose meta/info.json robot_type=Unitree_G1_Inspire (1035 episodes; 26-D state of G1 arm joints and Inspire hand joints). Dataset card does not say whether it is real or simulated; auto-generated model card." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/unitree-g1/xvla" }, { "robot": "ur5e", "policy": "act", "state": "works-with-caveats", "framework": "lerobot", "framework_version_tested": "0.5.1", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "Requires the UR5e-LeRobot extension (a LeRobot fork pinned at version 0.5.1, UR5e + PGI gripper), not an unmodified LeRobot install." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] }, "inference_real": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] }, "teleop_record": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] } }, "evidence": [ { "id": "ev-ur5e-lerobot-readme", "tier": "community-reported", "type": "source-code", "url": "https://github.com/yechen056/UR5e-LeRobot/blob/72d41bdae2606f464d3fd0da24cd6a20241ab244/README.md", "date": "2026-07-02", "verified_on": "2026-10-24", "versions": { "lerobot": "0.5.1" }, "summary": "Third-party UR5e LeRobot extension (listed on LeRobot's third-party robots page as a fork/extension) gives lerobot-train commands with --policy.type=act and real-robot lerobot-record eval commands for single-arm and bimanual UR5e, and states the policy was validated and tuned on its hardware.", "quote": "We have validated and tuned hyperparameters on this hardware stack for five policies: `ACT`, `Diffusion Policy`, `VQ-BeT`, `MultiTaskDiT`, `pi0`, and `pi05`." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/ur5e/act" }, { "robot": "ur5e", "policy": "diffusion", "state": "works-with-caveats", "framework": "lerobot", "framework_version_tested": "0.5.1", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "Requires the UR5e-LeRobot extension (a LeRobot fork pinned at version 0.5.1, UR5e + PGI gripper), not an unmodified LeRobot install." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] }, "inference_real": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] }, "teleop_record": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] } }, "evidence": [ { "id": "ev-ur5e-lerobot-readme", "tier": "community-reported", "type": "source-code", "url": "https://github.com/yechen056/UR5e-LeRobot/blob/72d41bdae2606f464d3fd0da24cd6a20241ab244/README.md", "date": "2026-07-02", "verified_on": "2026-10-24", "versions": { "lerobot": "0.5.1" }, "summary": "Third-party UR5e LeRobot extension (listed on LeRobot's third-party robots page as a fork/extension) gives lerobot-train commands with --policy.type=diffusion and real-robot lerobot-record eval commands for single-arm and bimanual UR5e, and states the policy was validated and tuned on its hardware.", "quote": "We have validated and tuned hyperparameters on this hardware stack for five policies: `ACT`, `Diffusion Policy`, `VQ-BeT`, `MultiTaskDiT`, `pi0`, and `pi05`." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/ur5e/diffusion" }, { "robot": "ur5e", "policy": "multitask-dit", "state": "works-with-caveats", "framework": "lerobot", "framework_version_tested": "0.5.1", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "Requires the UR5e-LeRobot extension (a LeRobot fork pinned at version 0.5.1, UR5e + PGI gripper), not an unmodified LeRobot install." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] }, "inference_real": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] }, "teleop_record": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] } }, "evidence": [ { "id": "ev-ur5e-lerobot-readme", "tier": "community-reported", "type": "source-code", "url": "https://github.com/yechen056/UR5e-LeRobot/blob/72d41bdae2606f464d3fd0da24cd6a20241ab244/README.md", "date": "2026-07-02", "verified_on": "2026-10-24", "versions": { "lerobot": "0.5.1" }, "summary": "Third-party UR5e LeRobot extension (listed on LeRobot's third-party robots page as a fork/extension) gives lerobot-train commands with --policy.type=multi_task_dit and real-robot lerobot-record eval commands for single-arm and bimanual UR5e, and states the policy was validated and tuned on its hardware.", "quote": "We have validated and tuned hyperparameters on this hardware stack for five policies: `ACT`, `Diffusion Policy`, `VQ-BeT`, `MultiTaskDiT`, `pi0`, and `pi05`." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/ur5e/multitask-dit" }, { "robot": "ur5e", "policy": "pi0-fast", "state": "works-with-caveats", "framework": "openpi", "caveats": [ "The evidence is the authors' own UR5e evaluation; openpi releases no UR5e π0-FAST checkpoint or UR5e training config." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-fast-paper" ] }, "inference_real": { "state": "works", "evidence": [ "ev-fast-paper" ] } }, "evidence": [ { "id": "ev-fast-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2501.09747", "date": "2025-01-16", "verified_on": "2026-10-24", "summary": "The FAST paper trains and evaluates π0-FAST on a single-arm UR5e table-bussing task and evaluates grocery bagging on a UR5e out of the box.", "quote": "This task requires a single UR5e robot arm to clean a table by bussing objects (a mixture of trash, plates, and dishes) into a trash can or bussing bin." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/ur5e/pi0-fast" }, { "robot": "ur5e", "policy": "pi0", "state": "works-with-caveats", "framework": "openpi", "caveats": [ "The upstream evidence is Physical Intelligence's own UR5e evaluation; openpi ships no UR5e fine-tuned checkpoint, and its examples/ur5 is an outline for writing UR5 input/output transforms rather than a tested recipe.", "Public end-to-end UR5e deployments are community-reported only (an openpi fork driven through a LeRobot plugin, and a LeRobot 0.5.1 fork)." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-pi0-paper", "ev-ur5e-gello-readme", "ev-ur5e-lerobot-readme" ] }, "inference_real": { "state": "works", "evidence": [ "ev-pi0-paper", "ev-ur5e-gello-readme", "ev-ur5e-lerobot-readme" ] } }, "evidence": [ { "id": "ev-pi0-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2410.24164", "date": "2024-10-31", "verified_on": "2026-10-24", "summary": "The π0 paper lists UR5e (one arm, two cameras, 7-D action space) among its platforms and fine-tunes and evaluates π0 on a UR5e bowl-stacking task.", "quote": "UR5e stack bowls. This task requires stacking bowls, with four bowls of different sizes." }, { "id": "ev-openpi-ur5-example", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/examples/ur5/README.md", "date": "2026-08-24", "verified_on": "2026-10-24", "summary": "openpi's UR5 example only outlines how to write UR5 input/output transforms for fine-tuning; it is not a released checkpoint or tested recipe.", "quote": "Below we provide an outline of how to implement the key components mentioned in the \"Finetune on your data\" section of the [README](../README.md) for finetuning on UR5 datasets." }, { "id": "ev-ur5e-gello-readme", "tier": "community-reported", "type": "source-code", "url": "https://github.com/F-Fer/lerobot_ur5e_gello/blob/3f3532cfc58d81ba12c1582f9c6087d12488269d/README.md", "date": "2026-02-16", "verified_on": "2026-10-24", "summary": "Third-party UR5e LeRobot plugin (Robotiq gripper, RTDE) records datasets with LeRobot and runs π0 checkpoints served by an openpi fork (openpi-ur5e, config pi0_ur5e) on the real UR5e, with an evaluation script.", "quote": "**Remote policy inference** — deploy OpenPI (π0) policies via WebSocket from a GPU server" }, { "id": "ev-ur5e-lerobot-readme", "tier": "community-reported", "type": "source-code", "url": "https://github.com/yechen056/UR5e-LeRobot/blob/72d41bdae2606f464d3fd0da24cd6a20241ab244/README.md", "date": "2026-07-02", "verified_on": "2026-10-24", "versions": { "lerobot": "0.5.1" }, "summary": "LeRobot route: the UR5e-LeRobot fork (lerobot 0.5.1) trains --policy.type=pi0 from lerobot/pi0_base and evaluates it on single-arm and bimanual UR5e.", "quote": "We have validated and tuned hyperparameters on this hardware stack for five policies: `ACT`, `Diffusion Policy`, `VQ-BeT`, `MultiTaskDiT`, `pi0`, and `pi05`." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/ur5e/pi0" }, { "robot": "ur5e", "policy": "pi05", "state": "works-with-caveats", "framework": "lerobot", "framework_version_tested": "0.5.1", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "Requires the UR5e-LeRobot extension (a LeRobot fork pinned at version 0.5.1, UR5e + PGI gripper), not an unmodified LeRobot install." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] }, "inference_real": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] }, "teleop_record": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] } }, "evidence": [ { "id": "ev-ur5e-lerobot-readme", "tier": "community-reported", "type": "source-code", "url": "https://github.com/yechen056/UR5e-LeRobot/blob/72d41bdae2606f464d3fd0da24cd6a20241ab244/README.md", "date": "2026-07-02", "verified_on": "2026-10-24", "versions": { "lerobot": "0.5.1" }, "summary": "Third-party UR5e LeRobot extension (listed on LeRobot's third-party robots page as a fork/extension) gives lerobot-train commands with --policy.type=pi05 and real-robot lerobot-record eval commands for single-arm and bimanual UR5e, and states the policy was validated and tuned on its hardware.", "quote": "We have validated and tuned hyperparameters on this hardware stack for five policies: `ACT`, `Diffusion Policy`, `VQ-BeT`, `MultiTaskDiT`, `pi0`, and `pi05`." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/ur5e/pi05" }, { "robot": "ur5e", "policy": "smolvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-tahsinop-smolvla", "ev-hub-leberkaes-smolvla" ] } }, "evidence": [ { "id": "ev-hub-tahsinop-smolvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/TahsinOP/smolvla_ur5e_insert_policy", "date": "2026-06-24", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config dataset kavish56345/insert_object_into_hole_20260623_v34 (32 episodes, overhead + wrist cameras), whose meta/info.json robot_type=ur5e. Auto-generated card, no rollout results." }, { "id": "ev-hub-leberkaes-smolvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Leberkaesweckle/Dataset_UR5e_Schunk_Gripper_50_episodes_10_09_2026_SMOLVLA", "date": "2026-09-11", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config dataset Leberkaesweckle/Dataset_UR5e_Schunk_Gripper_50_episodes_10_09_2026 (50 episodes), whose meta/info.json robot_type=ur5_schunk_follower. Auto-generated card, no rollout results." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/ur5e/smolvla" }, { "robot": "ur5e", "policy": "vqbet", "state": "works-with-caveats", "framework": "lerobot", "framework_version_tested": "0.5.1", "caveats": [ "Community-reported only; not documented upstream or maintainer-tested.", "Requires the UR5e-LeRobot extension (a LeRobot fork pinned at version 0.5.1, UR5e + PGI gripper), not an unmodified LeRobot install." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] }, "inference_real": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] }, "teleop_record": { "state": "works", "evidence": [ "ev-ur5e-lerobot-readme" ] } }, "evidence": [ { "id": "ev-ur5e-lerobot-readme", "tier": "community-reported", "type": "source-code", "url": "https://github.com/yechen056/UR5e-LeRobot/blob/72d41bdae2606f464d3fd0da24cd6a20241ab244/README.md", "date": "2026-07-02", "verified_on": "2026-10-24", "versions": { "lerobot": "0.5.1" }, "summary": "Third-party UR5e LeRobot extension (listed on LeRobot's third-party robots page as a fork/extension) gives lerobot-train commands with --policy.type=vqbet and real-robot lerobot-record eval commands for single-arm and bimanual UR5e, and states the policy was validated and tuned on its hardware.", "quote": "We have validated and tuned hyperparameters on this hardware stack for five policies: `ACT`, `Diffusion Policy`, `VQ-BeT`, `MultiTaskDiT`, `pi0`, and `pi05`." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/ur5e/vqbet" }, { "robot": "widowx-250", "policy": "act", "state": "partial", "partial_scope": "LeRobot training is evidenced by a public checkpoint. Real-robot ACT results on the WidowX 250 come only from the BridgeData V2 paper's own implementation, which was not released, not from LeRobot.", "framework": "lerobot", "caveats": [ "The BridgeData V2 evaluation used goal-image-conditioned ACT. bridge_data_v2 does not include the ACT implementation or a checkpoint." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-msnaga-act" ] }, "inference_real": { "state": "works-with-caveats", "notes": "BridgeData V2 authors' real-robot evaluation (10 trials per task), outside LeRobot.", "evidence": [ "ev-bridgev2-paper-act", "ev-bridgev2-repo" ] } }, "evidence": [ { "id": "ev-hub-msnaga-act", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/msnaga/widowx_250_pick_and_place_block_policy", "date": "2025-10-12", "verified_on": "2026-10-24", "summary": "config.json type=act. train_config dataset msnaga/widowx_250_pick_and_place_block, whose meta/info.json robot_type=widowx_250s_end_effector (50 episodes, one front camera, waist/shoulder/elbow/forearm_roll/wrist_angle/wrist_rotate/gripper joints). No model card or rollout results." }, { "id": "ev-bridgev2-paper-act", "tier": "community-reported", "type": "paper", "url": "https://arxiv.org/abs/2308.12952v3", "date": "2024-01-17", "verified_on": "2026-10-24", "summary": "BridgeData V2 paper (dataset paper; ACT first author Tony Zhao is a co-author) trains goal-conditioned ACT on BridgeData V2 and evaluates it on the real WidowX 250, 10 trials per task (Table 2).", "quote": "Action chunking with transformers (ACT). Previous work has also noted that predicting action sequences instead of single actions can benefit imitation learning [17, 18]." }, { "id": "ev-bridgev2-repo", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/rail-berkeley/bridge_data_v2/blob/bc60a35b701a12021c8c95e9d8601274d3acd928/README.md", "date": "2024-03-17", "verified_on": "2026-10-24", "summary": "bridge_data_v2 implements GCBC, D-GCBC, GCIQL, CRL and LCBC only. ACT is linked to its upstream repo, and no ACT checkpoint is released.", "quote": "We don't currently have a checkpoints for ACT or RT-1 available but may release them soon." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/widowx-250/act" }, { "robot": "widowx-250", "policy": "diffusion", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-catebros-diffusion" ] } }, "evidence": [ { "id": "ev-hub-catebros-diffusion", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/catebros/lerobot-widowx-diffusion", "date": "2026-01-10", "verified_on": "2026-10-24", "summary": "config.json type=diffusion. train_config dataset catebros/my_widowx_dataset, whose meta/info.json robot_type=w250_interbotix (31 episodes, two cameras, waist/shoulder/elbow/forearm_roll/wrist_angle/wrist_rotate/gripper joints). Auto-generated card, no rollout results." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/widowx-250/diffusion" }, { "robot": "widowx-250", "policy": "eo1", "state": "works", "framework": "eo1", "capabilities": { "train": { "state": "works", "evidence": [ "ev-eo1-paper-widowx" ] }, "inference_real": { "state": "works", "notes": "13-task 'WidowX Out-of-Box' kitchen suite plus open-world instruction following. The upstream client drives a WidowX 250s through the BridgeData WidowX controller stack with a RealSense D435.", "evidence": [ "ev-eo1-paper-widowx", "ev-eo1-repo-widowx" ] } }, "evidence": [ { "id": "ev-eo1-paper-widowx", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2508.21112v5", "date": "2026-02-25", "verified_on": "2026-10-24", "summary": "EO-1 paper (v5), Sec. 4.5: EO-1 is evaluated in the real world on a WidowX 250 S (13 out-of-box tasks) among other robots.", "quote": "We evaluate its performance on both short-horizon and long-horizon tasks using different robots, including Franka Panda, WidowX 250 S, Agibot G-1, and Lerobot SO100." }, { "id": "ev-eo1-repo-widowx", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/EO-Robotics/EO1/blob/451c363d7dff682700789afae0b9b17c004e8f8c/experiments/5_widowx/README.md", "date": "2025-11-12", "verified_on": "2026-10-24", "summary": "EO1 repo ships a WidowX 250s real-robot client (eval_widowx.py, bridge_data_robot submodule).", "quote": "This directory contains the implementation for controlling WidowX 250s robots using the EO-1 model." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/widowx-250/eo1" }, { "robot": "widowx-250", "policy": "groot-n1-7", "state": "partial", "partial_scope": "Simulation only: SimplerEnv WidowX + Bridge (N1.7 post-trained on BridgeData V2); no real-robot training or inference documented.", "framework": "isaac-groot", "capabilities": { "inference_sim": { "state": "works", "notes": "SimplerEnv Bridge (simulated WidowX-250) with nvidia/GR00T-N1.7-SimplerEnv-Bridge; NVIDIA reports N1.7 per-task success from 2% (eggplant in sink) to 100% (open drawer).", "evidence": [ "ev-groot-simplerenv-readme", "ev-groot-n17-bridge-card", "ev-simplerenv-paper" ] } }, "evidence": [ { "id": "ev-groot-simplerenv-readme", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/NVIDIA/Isaac-GR00T/blob/1a1837f20538b7d7e21f977a11a5aee14f99803c/examples/SimplerEnv/README.md", "date": "2026-07-07", "verified_on": "2026-10-25", "summary": "Isaac-GR00T SimplerEnv example: 'Bridge (WidowX robot)' section lists nvidia/GR00T-N1.7-SimplerEnv-Bridge, per-task N1.6/N1.7 success rates in SimplerEnv, and the fine-tuning command (--embodiment-tag SIMPLER_ENV_WIDOWX).", "quote": "| `widowx_spoon_on_towel` | 56/101 (55.4%) | 78/100 (78.0%) |" }, { "id": "ev-groot-n17-bridge-card", "tier": "upstream-documented", "type": "model-card", "url": "https://huggingface.co/nvidia/GR00T-N1.7-SimplerEnv-Bridge", "verified_on": "2026-10-25", "summary": "NVIDIA model card: GR00T N1.7 post-trained on the LeRobot conversion of BridgeData V2 for SimplerEnv.", "quote": "N1.7 post-trained model using the **Bridge Dataset** in SimplerEnv." }, { "id": "ev-simplerenv-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2405.05941v1", "date": "2024-05-09", "verified_on": "2026-10-25", "summary": "SimplerEnv paper: the simulated WidowX + Bridge setup models the WidowX-250 6DOF robot.", "quote": "For the WidowX + Bridge (with WidowX-250 6DOF robot), we adopt the following tasks:" } ], "tier": "upstream-documented", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/widowx-250/groot-n1-7" }, { "robot": "widowx-250", "policy": "openvla-oft", "state": "works-with-caveats", "framework": "openvla-oft", "caveats": [ "The WidowX result is an appendix experiment: OpenVLA-OFT (without FiLM) trained on all of BridgeData V2 and evaluated on a subset of OpenVLA's Bridge tasks (69.2% vs. 65.8% for OpenVLA).", "The openvla-oft repo ships fine-tuning and evaluation code only for LIBERO and ALOHA. It has no Bridge/WidowX config, deployment script or released WidowX checkpoint." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-oft-paper-bridge" ] }, "inference_real": { "state": "works", "evidence": [ "ev-oft-paper-bridge", "ev-bridgev2-paper" ] } }, "evidence": [ { "id": "ev-oft-paper-bridge", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2502.19645v2", "date": "2025-04-28", "verified_on": "2026-10-24", "summary": "OpenVLA-OFT paper, Appendix (Scaling up to BridgeData V2): OFT is trained on BridgeData V2 and evaluated on real WidowX tasks from OpenVLA's evaluation suite, scoring 69.2% on average vs. 65.8% for OpenVLA.", "quote": "Specifically, we train OpenVLA-OFT (without FiLM) on the BridgeData V2 dataset [52], which contains 50,365 real WidowX robot demonstrations (25× more than the four LIBERO task suites combined)." }, { "id": "ev-oft-readme", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/moojink/openvla-oft/blob/e4287e94541f459edc4feabc4e181f537cd569a8/README.md", "date": "2025-09-09", "verified_on": "2026-10-24", "summary": "openvla-oft README points to LIBERO and ALOHA for fine-tuning/evaluation. experiments/robot at this commit contains only aloha/ and libero/.", "quote": "See [ALOHA.md](ALOHA.md) for fine-tuning/evaluating on real-world ALOHA robot tasks." }, { "id": "ev-bridgev2-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2308.12952v3", "date": "2024-01-17", "verified_on": "2026-10-24", "summary": "BridgeData V2 was collected on a WidowX 250.", "quote": "showing the WidowX 250 robot arm and various cameras." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/widowx-250/openvla-oft" }, { "robot": "widowx-250", "policy": "openvla", "state": "works", "framework": "openvla", "capabilities": { "train": { "state": "works", "notes": "OpenVLA's pretraining mixture includes BridgeData V2, and the repo documents LoRA and full fine-tuning on BridgeData V2 (bridge_orig).", "evidence": [ "ev-openvla-readme-bridge-ft" ] }, "inference_real": { "state": "works", "notes": "Out-of-the-box evaluation on the real Bridge WidowX through the BridgeData V2 WidowX controller stack (bridge_data_robot, server-client).", "evidence": [ "ev-openvla-paper-widowx", "ev-openvla-readme-eval", "ev-bridgev2-paper" ] } }, "evidence": [ { "id": "ev-openvla-paper-widowx", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2406.09246v3", "date": "2024-09-05", "verified_on": "2026-10-24", "summary": "OpenVLA paper, Sec. 5.1: the authors evaluate OpenVLA out of the box on the real BridgeData V2 WidowX robot across visual, motion, physical, semantic and language-grounding tasks (Fig. 3, Table 4).", "quote": "We evaluate OpenVLA’s performance “out-of-the-box” on two robot embodiments: the WidowX robot from the BridgeData V2 evaluations [6] (see Fig. 1, left)" }, { "id": "ev-openvla-readme-eval", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/openvla/openvla/blob/c8f03f48af692657d3060c19588038c7220e9af9/README.md", "date": "2025-03-23", "verified_on": "2026-10-24", "summary": "OpenVLA README 'Evaluating OpenVLA' section: run real BridgeData V2 WidowX evals using the rail-berkeley/bridge_data_robot controller (widowx_envs) in Docker plus the openvla eval script.", "quote": "Added new section, [Evaluating OpenVLA](#evaluating-openvla), with instructions for running BridgeData V2 WidowX robot evals" }, { "id": "ev-openvla-readme-bridge-ft", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/openvla/openvla/blob/c8f03f48af692657d3060c19588038c7220e9af9/README.md", "date": "2025-03-23", "verified_on": "2026-10-24", "summary": "OpenVLA README gives LoRA and full fine-tuning commands on BridgeData V2.", "quote": "We show how you can fully fine-tune OpenVLA on [BridgeData V2]" }, { "id": "ev-bridgev2-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2308.12952v3", "date": "2024-01-17", "verified_on": "2026-10-24", "summary": "BridgeData V2 paper: the Bridge robot is a WidowX 250, which ties 'the WidowX robot from the BridgeData V2 evaluations' to this robot.", "quote": "showing the WidowX 250 robot arm and various cameras." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/widowx-250/openvla" }, { "robot": "widowx-250", "policy": "smolvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-asatheesh-smolvla-bridge" ] } }, "evidence": [ { "id": "ev-hub-asatheesh-smolvla-bridge", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/asatheesh/lerobot-smolvla-bridge", "date": "2025-08-29", "verified_on": "2026-10-25", "summary": "config.json type=smolvla (init lerobot/smolvla_base); train_config.json dataset IPEC-COMMUNITY/bridge_orig_lerobot (20000 steps), the LeRobot port of BridgeData V2 (bridge_orig) whose meta/info.json robot_type=widowx (53192 episodes, end-effector state/actions). Training on Bridge data only; no deployment on a WidowX shown." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/widowx-250/smolvla" }, { "robot": "widowx-250", "policy": "xvla", "state": "works-with-caveats", "framework": "x-vla", "caveats": [ "The paper says only 'WidowX' following the BridgeData V2 benchmark. It identifies the arm through BridgeData V2 (a WidowX 250), not by model name.", "X-VLA's repo provides an inference server but no WidowX robot client. Its released WidowX checkpoints (2toINF/X-VLA-WidowX, lerobot/xvla-widowx) are labelled for the SimplerEnv benchmark, and LeRobot has no WidowX 250 robot type." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-xvla-paper-widowx", "ev-xvla-readme-widowx" ] }, "inference_real": { "state": "works", "notes": "Real-world WidowX pick-and-place, five tasks (visual/motion/physical/semantic generalization), 10 trials each.", "evidence": [ "ev-xvla-paper-widowx", "ev-xvla-paper-realworld" ] }, "inference_sim": { "state": "works", "notes": "SimplerEnv WidowX (simulation).", "evidence": [ "ev-xvla-readme-widowx", "ev-lerobot-xvla-doc" ] } }, "evidence": [ { "id": "ev-xvla-paper-widowx", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2510.10274", "date": "2025-10-11", "verified_on": "2026-10-24", "summary": "X-VLA paper, Appendix J: X-VLA-0.9B fine-tuned on BridgeData is deployed on a real WidowX for pick-and-place, with each task evaluated 10 times.", "quote": "WidowX for pick-and-place experiments. Specifically, X-VLA-0.9B finetuned on BridgeData is directly deployed to evaluate its ability to perform robust manipulation on a compact platform." }, { "id": "ev-xvla-paper-realworld", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2510.10274", "date": "2025-10-11", "verified_on": "2026-10-24", "summary": "X-VLA paper, Sec. 5: real-world WidowX evaluation follows the BridgeData V2 benchmark (Fig. 7).", "quote": "Real-world experiments. We also evaluate X-VLA-0.9B on physical robotic platforms follow the BridgeDatav2 benchmark (Walke et al., 2023)" }, { "id": "ev-xvla-readme-widowx", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/2toinf/X-VLA/blob/6c4ecfb8be458ae7c0c96094e4b0add4174f4fe2/README.md", "date": "2026-10-20", "verified_on": "2026-10-24", "summary": "X-VLA README lists the 2toINF/X-VLA-WidowX checkpoint (fine-tuned on BridgeData V2, 95.8% on Simpler) and a server-client inference API.", "quote": "Fine-tuned on BridgeDataV2 (Simpler benchmark)." }, { "id": "ev-lerobot-xvla-doc", "tier": "upstream-documented", "type": "official-docs", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/docs/source/xvla.mdx", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot v0.6.1 X-VLA doc lists lerobot/xvla-widowx under 'Simulation Checkpoints'.", "quote": "Fine-tuned on BridgeData for pick-and-place experiments on compact WidowX platforms. Demonstrates robust manipulation capabilities." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/widowx-250/xvla" }, { "robot": "xarm", "policy": "act", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-jjbin-act", "ev-hub-shizhec-act" ] } }, "evidence": [ { "id": "ev-hub-jjbin-act", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/jjbin2001/act_xarm_black_white_sorting", "date": "2026-05-28", "verified_on": "2026-10-24", "summary": "config.json type=act; train_config dataset jjbin2001/xarm_black_white_sorting (100 episodes), whose meta/info.json robot_type=xarm7_follower. Auto-generated card, no rollout results." }, { "id": "ev-hub-shizhec-act", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Shizhec/xarm6_pick_toy_to_bowl_joint_act_v1", "date": "2026-05-14", "verified_on": "2026-10-24", "summary": "config.json type=act; train_config dataset Shizhec/xarm6_pick_toy_to_bowl_joint_v1 (100 episodes), whose meta/info.json robot_type='UFACTORY Robot' (xArm6). Auto-generated card, no rollout results." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/xarm/act" }, { "robot": "xarm", "policy": "diffusion", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-shizhec-diffusion" ] } }, "evidence": [ { "id": "ev-hub-shizhec-diffusion", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Shizhec/xarm6_pick_toy_to_bowl_joint_diffusion_v4", "date": "2026-09-14", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "config.json type=diffusion; train_config dataset Shizhec/xarm6_pick_toy_to_bowl_joint_v3, whose meta/info.json robot_type='UFACTORY Robot' (xArm6); card lists LeRobot version 0.6.1. Auto-generated card, no rollout results." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/xarm/diffusion" }, { "robot": "xarm", "policy": "evo1", "state": "works-with-caveats", "framework": "evo-1", "caveats": [ "Evidence covers the xArm6 only, driven by Evo-1's own websocket client script (official xArmAPI), not by the LeRobot xArm plugin; no xArm checkpoint is published." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-evo1-paper" ] }, "inference_real": { "state": "works", "evidence": [ "ev-evo1-paper", "ev-evo1-xarm6-client" ] } }, "evidence": [ { "id": "ev-evo1-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2511.04555", "date": "2025-11-06", "verified_on": "2026-10-24", "summary": "The Evo-1 paper's real-world experiments train and evaluate Evo-1 on a 6-DoF xArm6 (wrist + fixed camera) across four manipulation tasks.", "quote": "To evaluate the model’s performance in diverse real-world scenarios, we conduct experiments using a 6-DoF xArm6 robotic arm equipped with a parallel gripper, and design four manipulation tasks involving diverse object manipulation and real-time interaction" }, { "id": "ev-evo1-xarm6-client", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/MINT-SJTU/Evo-1/blob/5fd14b015013c4fd0aacf5f8f48f868ca9b870a2/README.md", "date": "2026-08-18", "verified_on": "2026-10-24", "summary": "The official Evo-1 repo ships an xArm6 inference client (Evo_1/scripts/Evo1_client_xarm6.py, using xarm.wrapper.XArmAPI).", "quote": "We provide an example of inference client script [Evo1_client_xarm6](Evo_1/scripts/Evo1_client_xarm6.py) for xArm6." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/xarm/evo1" }, { "robot": "xarm", "policy": "pi05", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-ys903-pi05" ] } }, "evidence": [ { "id": "ev-hub-ys903-pi05", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/ys903/pi05_xarm6_pick_place_50ep", "date": "2026-09-03", "verified_on": "2026-10-24", "summary": "config.json type=pi05 (pretrained_path lerobot/pi05_base); train_config dataset ys903/xarm6_pick_place_50_episodes (50 episodes), whose meta/info.json robot_type='UFACTORY Robot' (xArm6). Card lists LeRobot version 0.6.2, i.e. an unreleased main build. Auto-generated card, no rollout results." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/xarm/pi05" }, { "robot": "xarm", "policy": "smolvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "framework_version_tested": "0.6.1", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-zwonzis-smolvla", "ev-hub-jjbin-smolvla" ] } }, "evidence": [ { "id": "ev-hub-zwonzis-smolvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/zwonzis/smolvla-xarm7-pick-orange-ball", "date": "2026-07-27", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "config.json type=smolvla; train_config dataset zwonzis/xarm7-pick-orange-ball (56 episodes), whose meta/info.json robot_type=xarm7; card lists LeRobot version 0.6.1. Auto-generated card, no rollout results." }, { "id": "ev-hub-jjbin-smolvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/jjbin2001/smolvla_xarm_black_white_sorting", "date": "2026-05-27", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; train_config dataset jjbin2001/xarm_black_white_sorting, whose meta/info.json robot_type=xarm7_follower. Auto-generated card, no rollout results." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/xarm/smolvla" }, { "robot": "xarm", "policy": "tdmpc", "state": "works-with-caveats", "framework": "fowm", "caveats": [ "Evidence is the FOWM authors' real xArm 7 experiments with their own codebase; the public FOWM repo runs only the simulated xArm tasks, and LeRobot's tdmpc port documents no real-robot use.", "The real-robot results come from offline pretraining followed by online RL finetuning with task rewards (about 20 trials per task), not from pure imitation learning." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-fowm-paper" ] }, "inference_real": { "state": "works", "evidence": [ "ev-fowm-paper" ] } }, "evidence": [ { "id": "ev-fowm-paper", "tier": "upstream-documented", "type": "paper", "url": "https://arxiv.org/abs/2310.16029", "date": "2023-10-24", "verified_on": "2026-10-24", "summary": "FOWM (the method LeRobot's tdmpc policy implements) is pretrained offline and finetuned online on three visuo-motor tasks on a real xArm 7 with a jaw gripper and RealSense camera input.", "quote": "The agent controls an xArm 7 robot with a jaw gripper using positional control" }, { "id": "ev-lerobot-tdmpc-source", "tier": "upstream-documented", "type": "source-code", "url": "https://github.com/huggingface/lerobot/blob/v0.6.1/src/lerobot/policies/tdmpc/modeling_tdmpc.py", "date": "2026-08-03", "verified_on": "2026-10-24", "versions": { "lerobot": "0.6.1" }, "summary": "LeRobot's tdmpc policy is an implementation of FOWM, linking the grid policy to the FOWM paper's xArm results.", "quote": "Implementation of Finetuning Offline World Models in the Real World." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/xarm/tdmpc" }, { "robot": "xlerobot", "policy": "act", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "The documented ACT demo is single-arm: one XLeRobot arm is recorded and run as robot.type=so101_follower, not the full xlerobot robot type.", "XLeRobot is not a drop-in LeRobot plugin; dual-arm/VR recording requires copying XLeRobot code into a LeRobot checkout." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-xlerobot-act-doc" ] }, "inference_real": { "state": "works-with-caveats", "notes": "Docs show ACT running locally on the robot's Raspberry Pi.", "evidence": [ "ev-xlerobot-act-doc" ] } }, "evidence": [ { "id": "ev-xlerobot-act-doc", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/Vector-Wangel/XLeRobot/blob/b017b5e6354bd9f61f4247a920c72622ca0aade0/docs/en/source/software/getting_started/VLA_ACT.md", "verified_on": "2026-10-24", "summary": "XLeRobot project docs: ACT tutorial with a real-robot demo video (single arm, run on the Raspberry Pi), recording via robot.type=so101_follower.", "quote": "Single arm VLA policy implementation on XLeRobot:" } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/xlerobot/act" }, { "robot": "xlerobot", "policy": "diffusion", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-odog16-diffusion-tool", "ev-hub-odog16-diffusion-block" ] } }, "evidence": [ { "id": "ev-hub-odog16-diffusion-tool", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Odog16/tool_pickup_diffusion_1obs_policy", "date": "2026-02-28", "verified_on": "2026-10-25", "summary": "config.json type=diffusion; train_config.json dataset Odog16/tool_pickup (100000 steps), whose meta/info.json robot_type=xlerobot (100 episodes; two arms plus head, head + two wrist cameras)." }, { "id": "ev-hub-odog16-diffusion-block", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/Odog16/test_transfer_block_diffusion", "date": "2026-02-04", "verified_on": "2026-10-25", "summary": "config.json type=diffusion; train_config.json dataset Odog16/test_transfer_block (150000 steps), whose meta/info.json robot_type=xlerobot (154 episodes)." } ], "tier": "community-reported", "verified_on": "2026-10-25", "age_days": 0, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/xlerobot/diffusion" }, { "robot": "xlerobot", "policy": "pi05", "state": "works-with-caveats", "framework": "openpi", "caveats": [ "Community-reported only; not documented upstream by Physical Intelligence or maintainer-tested.", "Requires unmerged community forks: xuweiwu/XLeRobot (branch 'pi05 dual arm') and xuweiwu/openpi (branch 'biso101 training_support')." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-xlerobot-pi05-doc" ] }, "inference_real": { "state": "works-with-caveats", "evidence": [ "ev-xlerobot-pi05-doc" ] }, "teleop_record": { "state": "works", "evidence": [ "ev-xlerobot-pi05-dataset" ] } }, "evidence": [ { "id": "ev-xlerobot-pi05-doc", "tier": "community-reported", "type": "community-post", "url": "https://github.com/Vector-Wangel/XLeRobot/blob/b017b5e6354bd9f61f4247a920c72622ca0aade0/docs/en/source/software/getting_started/VLA_pi05.md", "verified_on": "2026-10-24", "summary": "XLeRobot docs page describing a community dual-arm pi0.5 implementation (openpi fork for fine-tuning, XLeRobot fork for teleop/recording/inference) with a demo video.", "quote": "This document provides information about the **pi0.5** VLA (Vision-Language-Action) implementation for XLeRobot with dual-arm support." }, { "id": "ev-xlerobot-pi05-dataset", "tier": "community-reported", "type": "community-post", "url": "https://huggingface.co/datasets/xuweiwu/bimanual-toy-box-cleanup", "verified_on": "2026-10-24", "summary": "The pi0.5 training dataset linked from the docs; meta/info.json robot_type=xlerobot, 43 episodes, LeRobot v3.0 format." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/xlerobot/pi05" }, { "robot": "xlerobot", "policy": "smolvla", "state": "works-with-caveats", "framework": "lerobot", "caveats": [ "The documented SmolVLA workflow drives XLeRobot's arms as robot.type=bi_so101_follower using a community LeRobot fork (kahowang/lerobot); XLeRobot is not a drop-in LeRobot plugin." ], "capabilities": { "train": { "state": "works", "evidence": [ "ev-xlerobot-smolvla-doc", "ev-hub-zonglin-smolvla", "ev-hub-monozu-smolvla" ] }, "inference_real": { "state": "works-with-caveats", "evidence": [ "ev-xlerobot-smolvla-doc" ] } }, "evidence": [ { "id": "ev-xlerobot-smolvla-doc", "tier": "upstream-documented", "type": "vendor-docs", "url": "https://github.com/Vector-Wangel/XLeRobot/blob/b017b5e6354bd9f61f4247a920c72622ca0aade0/docs/en/source/software/getting_started/VLA_smol.md", "verified_on": "2026-10-24", "summary": "XLeRobot project docs: SmolVLA trained on ~20 episodes with a bimanual SO-101 three-camera setup, with real-robot demo GIFs (drawer pick-place, zipper opening).", "quote": "After training on **~20 episodes**, XLeRobot can:" }, { "id": "ev-hub-zonglin-smolvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/zonglin11/smolvla-xlerobot-fold-towel", "date": "2026-06-14", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; trained on zonglin11/xlerobot_fold_towel whose meta/info.json robot_type=xlerobot_2wheels." }, { "id": "ev-hub-monozu-smolvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/monozu-deving/Xlerobot-grap-post-index4-v1", "date": "2026-03-25", "verified_on": "2026-10-24", "summary": "config.json type=smolvla; trained on monozu-deving/Xlerobot-grap-post-index4 whose meta/info.json robot_type=xlerobot_follower." } ], "tier": "upstream-documented", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/xlerobot/smolvla" }, { "robot": "xlerobot", "policy": "xvla", "state": "partial", "partial_scope": "Training evidenced by public checkpoint(s); on-robot inference not documented by a primary source.", "framework": "lerobot", "capabilities": { "train": { "state": "works", "evidence": [ "ev-hub-zonglin-xvla" ] } }, "evidence": [ { "id": "ev-hub-zonglin-xvla", "tier": "community-reported", "type": "hub-checkpoint", "url": "https://huggingface.co/zonglin11/xvla-xlerobot-fold", "date": "2026-06-07", "verified_on": "2026-10-24", "summary": "config.json type=xvla; trained on zonglin11/xlerobot_fold_cloth whose meta/info.json robot_type=xlerobot_2wheels." } ], "tier": "community-reported", "verified_on": "2026-10-24", "age_days": 1, "stale": false, "version_drift": false, "url": "https://caniembody.pages.dev/compat/xlerobot/xvla" } ], "inferred_untested": [ { "robot": "agilex-piper", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports AgileX PiPER through the third-party plugin lerobot_robot_wego_piper and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "agilex-piper", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports AgileX PiPER through the third-party plugin lerobot_robot_wego_piper and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "agilex-piper", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports AgileX PiPER through the third-party plugin lerobot_robot_wego_piper and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "agilex-piper", "policy": "groot-n1-7", "tier": "inferred", "basis": "Inferred: LeRobot supports AgileX PiPER through the third-party plugin lerobot_robot_wego_piper and GR00T N1.7 as --policy.type=groot since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "agilex-piper", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports AgileX PiPER through the third-party plugin lerobot_robot_wego_piper and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "agilex-piper", "policy": "molmoact2", "tier": "inferred", "basis": "Inferred: LeRobot supports AgileX PiPER through the third-party plugin lerobot_robot_wego_piper and MolmoAct2 as --policy.type=molmoact2 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "agilex-piper", "policy": "multitask-dit", "tier": "inferred", "basis": "Inferred: LeRobot supports AgileX PiPER through the third-party plugin lerobot_robot_wego_piper and Multitask DiT as --policy.type=multi_task_dit since 0.5.1, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "agilex-piper", "policy": "pi0-fast", "tier": "inferred", "basis": "Inferred: LeRobot supports AgileX PiPER through the third-party plugin lerobot_robot_wego_piper and π0-FAST as --policy.type=pi0_fast since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "agilex-piper", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports AgileX PiPER through the third-party plugin lerobot_robot_wego_piper and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "agilex-piper", "policy": "vla-jepa", "tier": "inferred", "basis": "Inferred: LeRobot supports AgileX PiPER through the third-party plugin lerobot_robot_wego_piper and VLA-JEPA as --policy.type=vla_jepa since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "groot-n1-7", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and GR00T N1.7 as --policy.type=groot since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "molmoact2", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and MolmoAct2 as --policy.type=molmoact2 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "multitask-dit", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and Multitask DiT as --policy.type=multi_task_dit since 0.5.1, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "pi0-fast", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and π0-FAST as --policy.type=pi0_fast since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "smolvla", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and SmolVLA as --policy.type=smolvla since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "vla-jepa", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and VLA-JEPA as --policy.type=vla_jepa since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "arx5", "policy": "xvla", "tier": "inferred", "basis": "Inferred: LeRobot supports ARX5 through the third-party plugin lerobot_robot_arx5 and X-VLA as --policy.type=xvla since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "bi-so", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports Bimanual SO-100/SO-101 natively since 0.3.2 and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "bi-so", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports Bimanual SO-100/SO-101 natively since 0.3.2 and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "bi-so", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports Bimanual SO-100/SO-101 natively since 0.3.2 and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "bi-so", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports Bimanual SO-100/SO-101 natively since 0.3.2 and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "bi-so", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports Bimanual SO-100/SO-101 natively since 0.3.2 and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "bi-so", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports Bimanual SO-100/SO-101 natively since 0.3.2 and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "bi-so", "policy": "vla-jepa", "tier": "inferred", "basis": "Inferred: LeRobot supports Bimanual SO-100/SO-101 natively since 0.3.2 and VLA-JEPA as --policy.type=vla_jepa since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "bi-so", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports Bimanual SO-100/SO-101 natively since 0.3.2 and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "bi-so", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports Bimanual SO-100/SO-101 natively since 0.3.2 and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "act", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and ACT as --policy.type=act, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "diffusion", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and Diffusion Policy as --policy.type=diffusion, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "groot-n1-7", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and GR00T N1.7 as --policy.type=groot since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "molmoact2", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and MolmoAct2 as --policy.type=molmoact2 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "multitask-dit", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and Multitask DiT as --policy.type=multi_task_dit since 0.5.1, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "pi0-fast", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and π0-FAST as --policy.type=pi0_fast since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "pi0", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and π0 as --policy.type=pi0 since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "pi05", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and π0.5 as --policy.type=pi05 since 0.4.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "smolvla", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and SmolVLA as --policy.type=smolvla since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "vla-jepa", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and VLA-JEPA as --policy.type=vla_jepa since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "earthrover-mini-plus", "policy": "xvla", "tier": "inferred", "basis": "Inferred: LeRobot supports EarthRover Mini Plus natively since 0.4.3 and X-VLA as --policy.type=xvla since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "diffusion", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and Diffusion Policy as --policy.type=diffusion, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "groot-n1-7", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and GR00T N1.7 as --policy.type=groot since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "molmoact2", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and MolmoAct2 as --policy.type=molmoact2 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "multitask-dit", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and Multitask DiT as --policy.type=multi_task_dit since 0.5.1, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "pi0-fast", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and π0-FAST as --policy.type=pi0_fast since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "pi0", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and π0 as --policy.type=pi0 since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "pi05", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and π0.5 as --policy.type=pi05 since 0.4.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "smolvla", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and SmolVLA as --policy.type=smolvla since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "vla-jepa", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and VLA-JEPA as --policy.type=vla_jepa since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "hope-jr", "policy": "xvla", "tier": "inferred", "basis": "Inferred: LeRobot supports HopeJR natively since 0.3.2 and X-VLA as --policy.type=xvla since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "i2rt-yam", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports I2RT YAM through the third-party plugin lerobot_robot_yam and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "i2rt-yam", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports I2RT YAM through the third-party plugin lerobot_robot_yam and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "i2rt-yam", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports I2RT YAM through the third-party plugin lerobot_robot_yam and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "i2rt-yam", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports I2RT YAM through the third-party plugin lerobot_robot_yam and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "i2rt-yam", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports I2RT YAM through the third-party plugin lerobot_robot_yam and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "i2rt-yam", "policy": "multitask-dit", "tier": "inferred", "basis": "Inferred: LeRobot supports I2RT YAM through the third-party plugin lerobot_robot_yam and Multitask DiT as --policy.type=multi_task_dit since 0.5.1, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "i2rt-yam", "policy": "pi0-fast", "tier": "inferred", "basis": "Inferred: LeRobot supports I2RT YAM through the third-party plugin lerobot_robot_yam and π0-FAST as --policy.type=pi0_fast since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "i2rt-yam", "policy": "pi0", "tier": "inferred", "basis": "Inferred: LeRobot supports I2RT YAM through the third-party plugin lerobot_robot_yam and π0 as --policy.type=pi0 since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "i2rt-yam", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports I2RT YAM through the third-party plugin lerobot_robot_yam and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "i2rt-yam", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports I2RT YAM through the third-party plugin lerobot_robot_yam and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "i2rt-yam", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports I2RT YAM through the third-party plugin lerobot_robot_yam and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "i2rt-yam", "policy": "xvla", "tier": "inferred", "basis": "Inferred: LeRobot supports I2RT YAM through the third-party plugin lerobot_robot_yam and X-VLA as --policy.type=xvla since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "koch", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports Koch v1.1 natively since 0.3.2 and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "koch", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports Koch v1.1 natively since 0.3.2 and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "koch", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports Koch v1.1 natively since 0.3.2 and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "koch", "policy": "groot-n1-7", "tier": "inferred", "basis": "Inferred: LeRobot supports Koch v1.1 natively since 0.3.2 and GR00T N1.7 as --policy.type=groot since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "koch", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports Koch v1.1 natively since 0.3.2 and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "koch", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports Koch v1.1 natively since 0.3.2 and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "koch", "policy": "molmoact2", "tier": "inferred", "basis": "Inferred: LeRobot supports Koch v1.1 natively since 0.3.2 and MolmoAct2 as --policy.type=molmoact2 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "koch", "policy": "multitask-dit", "tier": "inferred", "basis": "Inferred: LeRobot supports Koch v1.1 natively since 0.3.2 and Multitask DiT as --policy.type=multi_task_dit since 0.5.1, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "koch", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports Koch v1.1 natively since 0.3.2 and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "koch", "policy": "vla-jepa", "tier": "inferred", "basis": "Inferred: LeRobot supports Koch v1.1 natively since 0.3.2 and VLA-JEPA as --policy.type=vla_jepa since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "koch", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports Koch v1.1 natively since 0.3.2 and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "koch", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports Koch v1.1 natively since 0.3.2 and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "koch", "policy": "xvla", "tier": "inferred", "basis": "Inferred: LeRobot supports Koch v1.1 natively since 0.3.2 and X-VLA as --policy.type=xvla since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "groot-n1-7", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and GR00T N1.7 as --policy.type=groot since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "molmoact2", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and MolmoAct2 as --policy.type=molmoact2 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "multitask-dit", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and Multitask DiT as --policy.type=multi_task_dit since 0.5.1, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "pi0-fast", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and π0-FAST as --policy.type=pi0_fast since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "pi0", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and π0 as --policy.type=pi0 since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "vla-jepa", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and VLA-JEPA as --policy.type=vla_jepa since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "lekiwi", "policy": "xvla", "tier": "inferred", "basis": "Inferred: LeRobot supports LeKiwi natively since 0.3.2 and X-VLA as --policy.type=xvla since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "omx", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports ROBOTIS OMX natively since 0.4.3 and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "omx", "policy": "pi0-fast", "tier": "inferred", "basis": "Inferred: LeRobot supports ROBOTIS OMX natively since 0.4.3 and π0-FAST as --policy.type=pi0_fast since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "omx", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports ROBOTIS OMX natively since 0.4.3 and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "omx", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports ROBOTIS OMX natively since 0.4.3 and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "openarm", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports OpenArm natively since 0.4.4 and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "openarm", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports OpenArm natively since 0.4.4 and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "openarm", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports OpenArm natively since 0.4.4 and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "openarm", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports OpenArm natively since 0.4.4 and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "openarm", "policy": "multitask-dit", "tier": "inferred", "basis": "Inferred: LeRobot supports OpenArm natively since 0.4.4 and Multitask DiT as --policy.type=multi_task_dit since 0.5.1, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "openarm", "policy": "pi0-fast", "tier": "inferred", "basis": "Inferred: LeRobot supports OpenArm natively since 0.4.4 and π0-FAST as --policy.type=pi0_fast since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "openarm", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports OpenArm natively since 0.4.4 and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "openarm", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports OpenArm natively since 0.4.4 and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "diffusion", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and Diffusion Policy as --policy.type=diffusion, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "groot-n1-7", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and GR00T N1.7 as --policy.type=groot since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "molmoact2", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and MolmoAct2 as --policy.type=molmoact2 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "multitask-dit", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and Multitask DiT as --policy.type=multi_task_dit since 0.5.1, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "pi0-fast", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and π0-FAST as --policy.type=pi0_fast since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "vla-jepa", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and VLA-JEPA as --policy.type=vla_jepa since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "reachy2", "policy": "xvla", "tier": "inferred", "basis": "Inferred: LeRobot supports Reachy 2 natively since 0.4.0 and X-VLA as --policy.type=xvla since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "rebot-b601", "policy": "diffusion", "tier": "inferred", "basis": "Inferred: LeRobot supports reBot Arm B601-DM natively since 0.6.0 and Diffusion Policy as --policy.type=diffusion, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "rebot-b601", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports reBot Arm B601-DM natively since 0.6.0 and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "rebot-b601", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports reBot Arm B601-DM natively since 0.6.0 and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "rebot-b601", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports reBot Arm B601-DM natively since 0.6.0 and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "rebot-b601", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports reBot Arm B601-DM natively since 0.6.0 and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "rebot-b601", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports reBot Arm B601-DM natively since 0.6.0 and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "rebot-b601", "policy": "pi0-fast", "tier": "inferred", "basis": "Inferred: LeRobot supports reBot Arm B601-DM natively since 0.6.0 and π0-FAST as --policy.type=pi0_fast since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "rebot-b601", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports reBot Arm B601-DM natively since 0.6.0 and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "rebot-b601", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports reBot Arm B601-DM natively since 0.6.0 and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "rebot-b601", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports reBot Arm B601-DM natively since 0.6.0 and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "rebot-b601", "policy": "xvla", "tier": "inferred", "basis": "Inferred: LeRobot supports reBot Arm B601-DM natively since 0.6.0 and X-VLA as --policy.type=xvla since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "so100", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports SO-100 natively since 0.3.2 and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "so100", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports SO-100 natively since 0.3.2 and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "so100", "policy": "multitask-dit", "tier": "inferred", "basis": "Inferred: LeRobot supports SO-100 natively since 0.3.2 and Multitask DiT as --policy.type=multi_task_dit since 0.5.1, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "so100", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports SO-100 natively since 0.3.2 and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "so100", "policy": "vla-jepa", "tier": "inferred", "basis": "Inferred: LeRobot supports SO-100 natively since 0.3.2 and VLA-JEPA as --policy.type=vla_jepa since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "so100", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports SO-100 natively since 0.3.2 and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "so101", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports SO-101 natively since 0.3.2 and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "so101", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports SO-101 natively since 0.3.2 and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "so101", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports SO-101 natively since 0.3.2 and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "trossen-widowx", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports Trossen WidowX AI through the third-party plugin lerobot_robot_trossen and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "trossen-widowx", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports Trossen WidowX AI through the third-party plugin lerobot_robot_trossen and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "trossen-widowx", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports Trossen WidowX AI through the third-party plugin lerobot_robot_trossen and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "trossen-widowx", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports Trossen WidowX AI through the third-party plugin lerobot_robot_trossen and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "trossen-widowx", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports Trossen WidowX AI through the third-party plugin lerobot_robot_trossen and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "trossen-widowx", "policy": "molmoact2", "tier": "inferred", "basis": "Inferred: LeRobot supports Trossen WidowX AI through the third-party plugin lerobot_robot_trossen and MolmoAct2 as --policy.type=molmoact2 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "trossen-widowx", "policy": "multitask-dit", "tier": "inferred", "basis": "Inferred: LeRobot supports Trossen WidowX AI through the third-party plugin lerobot_robot_trossen and Multitask DiT as --policy.type=multi_task_dit since 0.5.1, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "trossen-widowx", "policy": "pi0-fast", "tier": "inferred", "basis": "Inferred: LeRobot supports Trossen WidowX AI through the third-party plugin lerobot_robot_trossen and π0-FAST as --policy.type=pi0_fast since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "trossen-widowx", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports Trossen WidowX AI through the third-party plugin lerobot_robot_trossen and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "trossen-widowx", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports Trossen WidowX AI through the third-party plugin lerobot_robot_trossen and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "trossen-widowx", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports Trossen WidowX AI through the third-party plugin lerobot_robot_trossen and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "unitree-g1", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports Unitree G1 natively since 0.4.3 and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "unitree-g1", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports Unitree G1 natively since 0.4.3 and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "unitree-g1", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports Unitree G1 natively since 0.4.3 and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "unitree-g1", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports Unitree G1 natively since 0.4.3 and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "unitree-g1", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports Unitree G1 natively since 0.4.3 and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "unitree-g1", "policy": "molmoact2", "tier": "inferred", "basis": "Inferred: LeRobot supports Unitree G1 natively since 0.4.3 and MolmoAct2 as --policy.type=molmoact2 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "unitree-g1", "policy": "multitask-dit", "tier": "inferred", "basis": "Inferred: LeRobot supports Unitree G1 natively since 0.4.3 and Multitask DiT as --policy.type=multi_task_dit since 0.5.1, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "unitree-g1", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports Unitree G1 natively since 0.4.3 and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "unitree-g1", "policy": "vla-jepa", "tier": "inferred", "basis": "Inferred: LeRobot supports Unitree G1 natively since 0.4.3 and VLA-JEPA as --policy.type=vla_jepa since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "unitree-g1", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports Unitree G1 natively since 0.4.3 and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "unitree-g1", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports Unitree G1 natively since 0.4.3 and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "ur5e", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports Universal Robots UR5e through the third-party plugin lerobot_robot_ur5e and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "ur5e", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports Universal Robots UR5e through the third-party plugin lerobot_robot_ur5e and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "ur5e", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports Universal Robots UR5e through the third-party plugin lerobot_robot_ur5e and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "ur5e", "policy": "groot-n1-7", "tier": "inferred", "basis": "Inferred: LeRobot supports Universal Robots UR5e through the third-party plugin lerobot_robot_ur5e and GR00T N1.7 as --policy.type=groot since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "ur5e", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports Universal Robots UR5e through the third-party plugin lerobot_robot_ur5e and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "ur5e", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports Universal Robots UR5e through the third-party plugin lerobot_robot_ur5e and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "ur5e", "policy": "molmoact2", "tier": "inferred", "basis": "Inferred: LeRobot supports Universal Robots UR5e through the third-party plugin lerobot_robot_ur5e and MolmoAct2 as --policy.type=molmoact2 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "ur5e", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports Universal Robots UR5e through the third-party plugin lerobot_robot_ur5e and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "ur5e", "policy": "vla-jepa", "tier": "inferred", "basis": "Inferred: LeRobot supports Universal Robots UR5e through the third-party plugin lerobot_robot_ur5e and VLA-JEPA as --policy.type=vla_jepa since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "ur5e", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports Universal Robots UR5e through the third-party plugin lerobot_robot_ur5e and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "ur5e", "policy": "xvla", "tier": "inferred", "basis": "Inferred: LeRobot supports Universal Robots UR5e through the third-party plugin lerobot_robot_ur5e and X-VLA as --policy.type=xvla since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xarm", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports UFACTORY xArm through the third-party plugin lerobot-robot-xarm and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xarm", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports UFACTORY xArm through the third-party plugin lerobot-robot-xarm and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xarm", "policy": "groot-n1-7", "tier": "inferred", "basis": "Inferred: LeRobot supports UFACTORY xArm through the third-party plugin lerobot-robot-xarm and GR00T N1.7 as --policy.type=groot since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xarm", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports UFACTORY xArm through the third-party plugin lerobot-robot-xarm and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xarm", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports UFACTORY xArm through the third-party plugin lerobot-robot-xarm and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xarm", "policy": "molmoact2", "tier": "inferred", "basis": "Inferred: LeRobot supports UFACTORY xArm through the third-party plugin lerobot-robot-xarm and MolmoAct2 as --policy.type=molmoact2 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xarm", "policy": "multitask-dit", "tier": "inferred", "basis": "Inferred: LeRobot supports UFACTORY xArm through the third-party plugin lerobot-robot-xarm and Multitask DiT as --policy.type=multi_task_dit since 0.5.1, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xarm", "policy": "pi0-fast", "tier": "inferred", "basis": "Inferred: LeRobot supports UFACTORY xArm through the third-party plugin lerobot-robot-xarm and π0-FAST as --policy.type=pi0_fast since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xarm", "policy": "pi0", "tier": "inferred", "basis": "Inferred: LeRobot supports UFACTORY xArm through the third-party plugin lerobot-robot-xarm and π0 as --policy.type=pi0 since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xarm", "policy": "vla-jepa", "tier": "inferred", "basis": "Inferred: LeRobot supports UFACTORY xArm through the third-party plugin lerobot-robot-xarm and VLA-JEPA as --policy.type=vla_jepa since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xarm", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports UFACTORY xArm through the third-party plugin lerobot-robot-xarm and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xarm", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports UFACTORY xArm through the third-party plugin lerobot-robot-xarm and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xarm", "policy": "xvla", "tier": "inferred", "basis": "Inferred: LeRobot supports UFACTORY xArm through the third-party plugin lerobot-robot-xarm and X-VLA as --policy.type=xvla since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xlerobot", "policy": "eo1", "tier": "inferred", "basis": "Inferred: LeRobot supports XLeRobot through the third-party plugin lerobot_robot_xlerobot and EO-1 as --policy.type=eo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xlerobot", "policy": "evo1", "tier": "inferred", "basis": "Inferred: LeRobot supports XLeRobot through the third-party plugin lerobot_robot_xlerobot and Evo-1 as --policy.type=evo1 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xlerobot", "policy": "fastwam", "tier": "inferred", "basis": "Inferred: LeRobot supports XLeRobot through the third-party plugin lerobot_robot_xlerobot and Fast-WAM as --policy.type=fastwam since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xlerobot", "policy": "groot-n1-7", "tier": "inferred", "basis": "Inferred: LeRobot supports XLeRobot through the third-party plugin lerobot_robot_xlerobot and GR00T N1.7 as --policy.type=groot since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xlerobot", "policy": "hil-serl", "tier": "inferred", "basis": "Inferred: LeRobot supports XLeRobot through the third-party plugin lerobot_robot_xlerobot and HIL-SERL as --policy.type=gaussian_actor since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xlerobot", "policy": "lingbot-va", "tier": "inferred", "basis": "Inferred: LeRobot supports XLeRobot through the third-party plugin lerobot_robot_xlerobot and LingBot-VA as --policy.type=lingbot_va since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xlerobot", "policy": "molmoact2", "tier": "inferred", "basis": "Inferred: LeRobot supports XLeRobot through the third-party plugin lerobot_robot_xlerobot and MolmoAct2 as --policy.type=molmoact2 since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xlerobot", "policy": "multitask-dit", "tier": "inferred", "basis": "Inferred: LeRobot supports XLeRobot through the third-party plugin lerobot_robot_xlerobot and Multitask DiT as --policy.type=multi_task_dit since 0.5.1, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xlerobot", "policy": "pi0-fast", "tier": "inferred", "basis": "Inferred: LeRobot supports XLeRobot through the third-party plugin lerobot_robot_xlerobot and π0-FAST as --policy.type=pi0_fast since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xlerobot", "policy": "pi0", "tier": "inferred", "basis": "Inferred: LeRobot supports XLeRobot through the third-party plugin lerobot_robot_xlerobot and π0 as --policy.type=pi0 since 0.3.2, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xlerobot", "policy": "tdmpc", "tier": "inferred", "basis": "Inferred: LeRobot supports XLeRobot through the third-party plugin lerobot_robot_xlerobot and TD-MPC as --policy.type=tdmpc, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xlerobot", "policy": "vla-jepa", "tier": "inferred", "basis": "Inferred: LeRobot supports XLeRobot through the third-party plugin lerobot_robot_xlerobot and VLA-JEPA as --policy.type=vla_jepa since 0.6.0, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xlerobot", "policy": "vqbet", "tier": "inferred", "basis": "Inferred: LeRobot supports XLeRobot through the third-party plugin lerobot_robot_xlerobot and VQ-BeT as --policy.type=vqbet, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." }, { "robot": "xlerobot", "policy": "wall-oss", "tier": "inferred", "basis": "Inferred: LeRobot supports XLeRobot through the third-party plugin lerobot_robot_xlerobot and WALL-OSS as --policy.type=wall_x since 0.4.3, so its tools connect them. No run has been published. Check action and state dimensions, camera count and compute before relying on it." } ] } }