sock-heist-6m-continue-2-e
Another checkpoint of my second sock heist run. It slips past the guards and grabs the sock without getting caught. It finishes every run in under 10 seconds.
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> GinDuck/microduck-sock-heist-6m-continue-2-e
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-SockHeist6m-MicroDuck - repo:
https://github.com/pollen-robotics/microduck-challenges - branch:
sock-continue-2 - commit:
10ba3a426 - checkpoint:
600 - seed:
1 - base:
mjlab-microduck 0.1.0 @ 1efd853be - started:
2026-11-01T07:12:53Z
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
cd microduck-challenges
git checkout 10ba3a426
uv sync
uv run train Mjlab-SockHeist6m-MicroDuck --env.scene.num-envs 4096 --agent.max-iterations 1000 --agent.seed 1 --agent.logger tensorboard --agent.run-name continue-2
- Downloads last month
- 14