Does π0.5 run on Trossen WidowX AI?
Capabilities
| Capability | State | Notes | Evidence |
|---|---|---|---|
| Train / fine-tune | works | ev-trossen-pi05-lerobot-card, ev-trossen-pi05-pour-card | |
| Real-robot inference | works | ev-trossen-plugin-readme, ev-trossen-pi05-lerobot-card, ev-trossen-pi05-pour-card | |
| Simulation inference | no verified evidence | — | |
| Teleop + record | no verified evidence | — |
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).
Evidence 4 sources; the cell shows the highest tier
- T1 · upstream documentedvendor-docssource 2026-09-16verified 2026-10-24 (1d ago)
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.
“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).”
https://github.com/TrossenRobotics/lerobot_trossen/blob/337ed74df1ee617ddfb5987a9f7464cf3089c4a2/README.md - T1 · upstream documentedmodel-cardsource 2026-06-03verified 2026-10-24 (1d ago)
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.
“π₀.₅ fine-tune for bimanual red-cube handover / block transfer on the Trossen AI”
https://huggingface.co/TrossenRoboticsCommunity/pi05-block-transfer-lerobot/blob/ef188f51fd55da409f315472fe1e91e457bb6162/README.md - T1 · upstream documentedmodel-cardsource 2026-09-25verified 2026-10-24 (1d ago)
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.
“Evaluated on the real robot — the bimanual WidowX AI stationary rig — and reported by the operator as performing the pouring task well.”
https://huggingface.co/TrossenRoboticsCommunity/bimanual-widowxai-pour-items-red-cup-pi05/blob/e4281ebd9506852c8f9526ba3553181e4651a853/README.md - T1 · upstream documentedvendor-docsverified 2026-10-24 (1d ago)
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.
“openpi_trossen_ai_pi05_red_block [success] : π0.5 model successfully picks up and transfers red block on first try.”
https://docs.trossenrobotics.com/trossen_arm/v1.9/tutorials/openpi.html
π0.5 inference speed by hardware robot-independent · inference cost only
| Compute | Runtime | Precision | Cams | p50 / p95 ms | Inference | Platform | Source |
|---|---|---|---|---|---|---|---|
| RTX 4090 | llama.cpp | fp16 | 2 @224x224 | 51.7 / — | 19.34 Hz | —conditionsboard: NVIDIA RTX 4090 · ggml_cuda_f16_f32_out: 0 (README preset for RTX 4090) Inference cost only; not task success. Full VLM + action-expert pass (ViT + LLM + 10-step action expert), LIBERO inputs (agent-view + wrist at 224x224). From the EagleVLA-Edge README RTX 4090 table (P50), PR #6 configuration. The README's naive π0.5 baseline on this GPU: 162.9 ms (not filed as a separate run: the baseline precision is not stated). | T3 · community reported |
| RTX 4090 | llama.cpp | fp16 | 2 @224x224 | 58.5 / — | 17.09 Hz | —conditionsboard: NVIDIA RTX 4090 · ggml_cuda_f16_f32_out: 0 (README preset for RTX 4090) Inference cost only; not task success. Full VLM + action-expert pass (ViT + LLM + 10-step action expert), LIBERO inputs (agent-view + wrist at 224x224). From the EagleVLA-Edge README RTX 4090 table (P50). The README's naive π0.5 baseline on this GPU: 162.9 ms (not filed as a separate run: the baseline precision is not stated). | T3 · community reported |
| Jetson AGX Thor 128GB | llama.cpp | fp16 | 2 @224x224 | 132.8 / — | 7.53 Hz | —conditionsboard: Jetson Thor (paper/README: "Jetson Thor") Inference cost only; not task success. Full VLM + action-expert pass (ViT + LLM + 10-step action expert), LIBERO inputs (agent-view + wrist at 224x224). Later configuration from the EagleVLA-Edge README (PR #6); the README notes an accuracy caveat for the two-switch GEMM preset on Thor. Naive PyTorch π0.5 on the same board per the paper: 457.9 ms (2.18 Hz) (not filed as a separate run: the baseline precision is not stated). | T3 · community reported |
| Jetson AGX Thor 128GB | llama.cpp | fp16 | 2 @224x224 | 202.8 / — | 4.93 Hz | —conditionsboard: Jetson Thor (paper/README: "Jetson Thor") Inference cost only; not task success. Full VLM + action-expert pass (ViT + LLM + 10-step action expert), LIBERO inputs (agent-view + wrist at 224x224). Paper's 'control frequency' (10.25 Hz) is 1/reaction time with VLM-skipping scheduling; not executed actions/s. Board inferred to be the 128 GB T5000 class from the paper's 130 W max-power entry for Thor (T4000 is 40–70 W). Naive PyTorch π0.5 on the same board per the paper: 457.9 ms (2.18 Hz) (not filed as a separate run: the baseline precision is not stated). | T3 · community reported |
| Jetson AGX Orin 64GB | llama.cpp | fp16 | 2 @224x224 | 271.1 / — | 3.69 Hz | power_mode 50W (paper label "Orin-50W")conditionsboard: Jetson Orin, 'Orin-50W' power mode (AGX Orin 64GB inferred, see ev-orin-board-inferred) Inference cost only; not task success. Full VLM + action-expert pass (ViT + LLM + 10-step action expert), LIBERO inputs (agent-view + wrist at 224x224). Later configuration from the EagleVLA-Edge README (time cache + cuBLASLt GEMM tuning, PR #6); not in the paper. Naive PyTorch π0.5 on the same board per the paper: 1420.8 ms (0.70 Hz) (not filed as a separate run: the baseline precision is not stated). | T3 · community reported |
| Jetson AGX Orin 64GB | llama.cpp | fp16 | 2 @224x224 | 374.5 / — | 2.67 Hz | power_mode 50W (paper label "Orin-50W")conditionsboard: Jetson Orin, 'Orin-50W' power mode (AGX Orin 64GB inferred, see ev-orin-board-inferred) Inference cost only; not task success. Full VLM + action-expert pass (ViT + LLM + 10-step action expert), LIBERO inputs (agent-view + wrist at 224x224). Paper's 'control frequency' (6.93 Hz) is 1/reaction time with confidence-based scheduling that skips the VLM; it is not executed actions/s and is not recorded as control_rate_hz. Naive PyTorch π0.5 on the same board per the paper: 1420.8 ms (0.70 Hz) (not filed as a separate run: the baseline precision is not stated). | T3 · community reported |
| Jetson Orin Nano 8GB (Super) | TensorRT | fp16 | 2 @640x480 | 438.47 / 438.93 | 2.28 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; 10 denoising steps; latency is runtime-only over a pre-materialized observation ring. Whole-board RAM in use 6.39 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) | TensorRT | fp16 | 3 @640x480 | 643.39 / 644.92 | 1.55 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; 10 denoising steps; latency is runtime-only over a pre-materialized observation ring. Whole-board RAM in use 6.18 GB (unified memory, desktop running). Actions were gated against the stock policy (cosine >= 0.99999) before timing. | T3 · community reported |
Embed this cell
Badge for READMEs and model cards. It always shows the current state.
[](https://caniembody.pages.dev/compat/trossen-widowx/pi05)<iframe src="https://caniembody.pages.dev/embed/trossen-widowx/pi05" width="380" height="132" style="border:0" title="π0.5 on Trossen WidowX AI | CanIEmbody"></iframe>History
- 2026-10-25 fecccd1 Initial release: robot policy compatibility matrix · Danilo Znamerovszkij