got-a-talent-dance2
Inspired by Billy Ray Cyrus and Achy Breaky Heart, MicroDuck hits the floor for the Electric Slide: vine right, vine left, back it up, rock and turn, then again from the top. Twenty seconds, two quarter turns, not one tumble. Its heart may ache, but its feet sure don’t!
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> TheSauceman/microduck-got-a-talent-dance2
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-GotATalent-MicroDuck - repo:
https://github.com/SteveTarter/microduck-challenges - branch:
claude/project-thread-h4e0o0 - commit:
238700562 - checkpoint:
5999 - seed:
1 - base:
mjlab-microduck 0.1.0 @ 8d0dbc94c - started:
2026-10-19T21:55:47Z
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/SteveTarter/microduck-challenges
cd microduck-challenges
git checkout 238700562
uv sync
uv run train Mjlab-GotATalent-MicroDuck --env.scene.num-envs 2048 --agent.max-iterations 6000 --agent.seed 1 --agent.logger tensorboard --agent.run-name dance2
- Downloads last month
- 27