Does π0-FAST run on Franka Emika Panda?
works with caveats
Capabilities
| Capability | State | Notes | Evidence |
|---|---|---|---|
| Train / fine-tune | works | openpi documents training generalist policies on the full DROID dataset (approximate open-source pipeline for pi0-FAST-DROID). | ev-openpi-readme-fast-droid, ev-fast-paper-droid |
| Real-robot inference | works | ev-fast-paper-droid, ev-openpi-readme-fast-droid, ev-openpi-droid-example, ev-droid-paper | |
| Simulation inference | no verified evidence | — | |
| Teleop + record | no verified evidence | — |
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.
Evidence 4 sources; the cell shows the highest tier
- T1 · upstream documentedpapersource 2025-01-16verified 2026-10-24 (1d ago)
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.
“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.”
https://arxiv.org/abs/2501.09747 - T1 · upstream documentedofficial-docssource 2026-08-24verified 2026-10-24 (1d ago)
openpi lists the pi0_fast_droid 'Inference' checkpoint for 0-shot use on the DROID robot platform.
“$\pi_0$-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”
https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/README.md - T1 · upstream documentedofficial-docssource 2026-08-24verified 2026-10-24 (1d ago)
openpi DROID example: serve pi0_fast_droid and run the DROID robot client (main.py) on the DROID control laptop.
“uv run scripts/serve_policy.py policy:checkpoint --policy.config=pi0_fast_droid --policy.dir=gs://openpi-assets/checkpoints/pi0_fast_droid”
https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/examples/droid/README.md - T1 · upstream documentedpapersource 2025-04-22verified 2026-10-24 (1d ago)
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.)
“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
Embed this cell
Badge for READMEs and model cards. It always shows the current state.
[](https://caniembody.pages.dev/compat/franka-panda/pi0-fast)<iframe src="https://caniembody.pages.dev/embed/franka-panda/pi0-fast" width="380" height="132" style="border:0" title="π0-FAST on Franka Emika Panda | CanIEmbody"></iframe>History
- 2026-10-25 fecccd1 Initial release: robot policy compatibility matrix · Danilo Znamerovszkij