Does π0 run on Franka Emika Panda?

Capabilities

Caveats

• openpi lists pi0_droid as a fine-tuning starting point and says it may not follow language commands as well as π0-FAST-DROID. On real DROID tasks PI reports a lower score for π0 (0.49) than for its knowledge-insulated model (0.55).
• Documented for the DROID platform (Panda + ZED cameras + DROID control stack) through a remote policy server, not for an arbitrary Panda setup.

Evidence 4 sources; the cell shows the highest tier

1. T1 · upstream documentedpapersource 2025-05-29verified 2026-10-24 (1d ago)

   Physical Intelligence's knowledge-insulation paper evaluates π0 (and its own method) on the real-world DROID benchmark in unseen scenes. π0 scored 0.49 ± 0.09 task progress.

   “We also evaluate our generalist on the open source benchmark DROID [23] for the same set of tabletop manipulation tasks as in [37]. Our method received a score of 0.55 ± 0.09, π0 received 0.49 ± 0.09”

   https://arxiv.org/abs/2505.23705
2. T1 · upstream documentedofficial-docssource 2026-08-24verified 2026-10-24 (1d ago)

   openpi lists the pi0_droid checkpoint (usage: Fine-Tuning).

   “$\pi_0$ model fine-tuned on the [DROID dataset](https://droid-dataset.github.io/): faster inference than $\pi_0$-FAST-DROID, but may not follow language commands as well”

   https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/README.md
3. T1 · upstream documentedofficial-docssource 2026-08-24verified 2026-10-24 (1d ago)

   openpi DROID example gives a server config for pi0_droid, then the same DROID robot client as for π0.5.

   “uv run scripts/serve_policy.py policy:checkpoint --policy.config=pi0_droid --policy.dir=gs://openpi-assets/checkpoints/pi0_droid”

   https://github.com/Physical-Intelligence/openpi/blob/215abfb217dbac7d5f1273282331b9b1866c0479/examples/droid/README.md
4. 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