Does SmolVLA run on Reachy 2?
Capabilities
| Capability | State | Notes | Evidence |
|---|---|---|---|
| Train / fine-tune | works | ev-hub-erlhub-smolvla-reachy2, ev-hub-erlhub-smolvla-ft-reachy2 | |
| Real-robot inference | works with caveats | Community deployment via lerobot-record --policy.path on a physical Reachy 2; scored 6.5 on an undefined /10 scale. | ev-reachy-pollen-smolvla-deploy |
| Simulation inference | no verified evidence | — | |
| Teleop + record | no verified evidence | — |
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.
Evidence 3 sources; the cell shows the highest tier
- T3 · community reportedhub-checkpointsource 2026-04-28verified 2026-10-25 (today)
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.
https://huggingface.co/erl-hub/reachy-smolvla-ft-execute-pick-and-place - T3 · community reportedsource-codesource 2026-08-21verified 2026-10-25 (today)
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)'.
“| `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 |”
https://github.com/NikolaRaicevic2001/reachy_pollen/blob/7edb3a080d1cd68d88a0f03a518589260e9110dd/README.md - T3 · community reportedhub-checkpointverified 2026-10-24 (1d ago)
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).
https://huggingface.co/erl-hub/reachy-smolvla
SmolVLA inference speed by hardware robot-independent · inference cost only
| Compute | Runtime | Precision | Cams | p50 / p95 ms | Inference | Platform | Source |
|---|---|---|---|---|---|---|---|
| Jetson Orin Nano 8GB (Super) | TensorRT | fp16 | 2 @640x480 | 107.55 / 108.48 | 9.3 Hz | jetpack 7.2.1 · power_mode MAXN_SUPER, clocks pinned with jetson_clocks (GPU 1020 MHz, CPU 1728 MHz) · tensorrt 10.16.2.10conditionsboard: Jetson Orin Nano Engineering Reference Developer Kit Super (8 GB) · l4t: R39.2.1 · os: Ubuntu 24.04.4 LTS · python: 3.12.3 · desktop: running Inference cost only; not task success. 300 s run on synthetic 640x480 observations resized by the exported policy to 512x512; 10 denoising steps; latency is runtime-only over a pre-materialized observation ring. Whole-board RAM in use 1.78 GB (unified memory, desktop running). Actions were gated against the stock policy (cosine >= 0.99999) before timing. | T3 · community reported |
| Jetson Orin Nano 8GB (Super) | ONNX (split) | fp16 | 2 | 189.89 / 190.93 | 5.25 Hz | jetpack 7.2 (L4T R39.2.x, per the repo host-setup doc at this commit) · power_mode MAXN_SUPER, pinned clocks · cuda 13.2 · tensorrt 10.16.2 (system, via ONNX Runtime TensorRT EP)conditionsboard: Jetson Orin Nano Super 8 GB developer kit · os: Ubuntu 24.04 · onnxruntime-gpu: 1.24 Inference cost only; not task success. Deterministic synthetic observations, pinned MAXN_SUPER clocks. Configured EP priority 7 graphs TensorRT / 2 CPU. System RAM 2385 MB in use. | T3 · community reported |
| Jetson Orin Nano 8GB (Super) | PyTorch eager | fp32 | 2 | 1167.93 / 1176.65 | 0.86 Hz | jetpack 7.2 (L4T R39.2.x, per the repo host-setup doc at this commit) · power_mode MAXN_SUPER, pinned clocks · cuda 13.2 · lerobot 0.5.1conditionsboard: Jetson Orin Nano Super 8 GB developer kit · os: Ubuntu 24.04 Inference cost only; not task success. Deterministic synthetic observations, pinned MAXN_SUPER clocks. Stock LeRobot SmolVLAPolicy in PyTorch FP32 (LeRobot 0.5.1 env, JetPack-matched torch). RSS peak 3826 MB, system RAM 4126 MB in use. | T3 · community reported |
Embed this cell
Badge for READMEs and model cards. It always shows the current state.
[](https://caniembody.pages.dev/compat/reachy2/smolvla)<iframe src="https://caniembody.pages.dev/embed/reachy2/smolvla" width="380" height="132" style="border:0" title="SmolVLA on Reachy 2 | CanIEmbody"></iframe>