SccHy/microduck-sprint-hc901

sprint-hc901

sprint-hc901

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

Run it on a robot

sudo robotctl policy load <slot> SccHy/microduck-sprint-hc901

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

• repo: pollen-robotics/microduck_rl
• branch: challenge-recipe-publishing
• commit: 1efd853be

2 m Sprint — 0.901s

Official Arena run (5/5 seeds, mean 0.901s, ~69th of 249). The duck ignores the 1.2 m/s command cap and sprints at ~2.2 m/s with a 6.7 Hz bounce gait — watch the head counter-swing like a sprinter's arms:

Acknowledgments

This policy is a hill-climbing fine-tune of a distilled pure-MLP variant derived from a1rb4Ck/microduck-sprint2m (Apache-2.0), which itself builds on the namenu → juicenv → suraj-phanindra remix chain. Thanks to all original authors — this model would not exist without their work.