Does π0.5 run on Franka Emika Panda?

works with caveats

T1 · upstream documentedlast verified 2026-10-24 · 1d agoopenpi

Capabilities

Capability	State	Notes	Evidence
Train / fine-tune	works	openpi documents fine-tuning π0.5 on a custom DROID dataset.	ev-openpi-readme-pi05-droid
Real-robot inference	works		ev-openpi-droid-example, ev-ki-paper-droid, ev-droid-paper
Simulation inference	no verified evidence		—
Teleop + record	no verified evidence		—

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.

Evidence 4 sources; the cell shows the highest tier

T1 · upstream documentedofficial-docssource 2026-08-24verified 2026-10-24 (1d ago)
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.
“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).”
https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/examples/droid/README.md
T1 · upstream documentedofficial-docssource 2026-08-24verified 2026-10-24 (1d ago)
openpi lists pi05_droid (usage: Inference / Fine-Tuning), a π0.5 fine-tuned on DROID with knowledge insulation.
“$\pi_{0.5}$ 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”
https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/README.md
T1 · upstream documentedpapersource 2025-05-29verified 2026-10-24 (1d ago)
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.
“We evaluate our method on the simulated LIBERO [30] benchmark, as well as the real world DROID [23] benchmark.”
https://arxiv.org/abs/2505.23705
T1 · upstream documentedpapersource 2025-04-22verified 2026-10-24 (1d ago)
The DROID platform and dataset are built on the Franka Emika Panda.
“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.”
https://arxiv.org/abs/2403.12945v2

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	— conditions board: 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	— conditions board: 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	— conditions board: 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	— conditions board: 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") conditions board: 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") conditions board: 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 10.16.2.10 conditions board: 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 10.16.2.10 conditions board: 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.

[![π0.5 on Franka Emika Panda: works with caveats](https://caniembody.pages.dev/badge/franka-panda/pi05.svg)](https://caniembody.pages.dev/compat/franka-panda/pi05)

<iframe src="https://caniembody.pages.dev/embed/franka-panda/pi05" width="380" height="132" style="border:0" title="π0.5 on Franka Emika Panda | CanIEmbody"></iframe>

History

2026-10-25 fecccd1 Initial release: robot policy compatibility matrix · Danilo Znamerovszkij