Does ACT run on ALOHA (stationary)?

Capabilities

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.

Evidence 5 sources; the cell shows the highest tier

1. T1 · upstream documentedpapersource 2023-04-23verified 2026-10-24 (1d ago)

   The ACT paper introduces both ALOHA and ACT, and evaluates ACT on 6 real-world ALOHA tasks plus 2 simulated (MuJoCo) ALOHA tasks.

   “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.”

   https://arxiv.org/abs/2304.13705
2. T1 · upstream documentedofficial-docsverified 2026-10-24 (1d ago)

   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.

   “This tutorial explains how to use [Aloha and Aloha 2 stationary](https://www.trossenrobotics.com/aloha-stationary) with LeRobot.”

   https://github.com/huggingface/lerobot/blob/bed90e3a41c43758c619dba66158ddd5798d361a/examples/9_use_aloha.md
3. T1 · upstream documentedvendor-docsverified 2026-10-24 (1d ago)

   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.

   “The policy is specified with policy=act_aloha_real.”

   https://docs.trossenrobotics.com/aloha_docs/2.0/training/lerobot_guide.html
4. T3 · community reportedhub-checkpointverified 2026-10-24 (1d ago)

   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.

   https://huggingface.co/TrossenRoboticsCommunity/act-fold-tshirt-aloha
5. T1 · upstream documentedmodel-cardverified 2026-10-24 (1d ago)

   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).

   “trained for the `AlohaTransferCube` environment from [gym-aloha](https://github.com/huggingface/gym-aloha).”

   https://huggingface.co/lerobot/act_aloha_sim_transfer_cube_human