sprint-2m-reference

Pollen's reference for the Microduck Arena's 2 m Sprint: the sprint_2m challenge of microduck-challenges trained unchanged with its recipe (4096 envs, 3000 iterations, seed 1); checkpoint 1000, its fastest on the Arena.

A perpetual policy for the microduck (61-D observation, 14 actions, 50 Hz). Runs until told otherwise — a gait for the walk slot.

Run it on a robot

sudo robotctl policy load walk pollen-robotics/microduck-sprint-2m-reference

The observation normalizer is baked into policy.onnx; feed raw observations. manifest.json follows schema 2 of the microduck policy manifest (docs/policy-manifest.md in the daemon repo).

Training

  • task_id: Mjlab-Sprint2m-MicroDuck
  • repo: https://github.com/pollen-robotics/microduck-challenges.git
  • branch: heading-hold
  • commit: 7a668f11c
  • checkpoint: 1000
  • seed: 1
  • base: mjlab-microduck 0.1.0 @ 981a279c6
  • started: 2026-09-28T15:04:10Z

Reproduce

Same code, same uv.lock, same command, same seed. Training it again yields a comparable policy, not the same weights: GPU reinforcement learning is not bit-reproducible across machines.

git clone https://github.com/pollen-robotics/microduck-challenges.git
cd microduck-challenges
git checkout 7a668f11c
uv sync
uv run train Mjlab-Sprint2m-MicroDuck --env.scene.num-envs 4096 --agent.max-iterations 3000 --agent.seed 1 --agent.logger tensorboard --agent.run-name reference
Downloads last month

-

Downloads are not tracked for this model. How to track
Video Preview
loading