bubu-hill-robust
Own lineage, built to be fast and stable off the official seeds: our dive1 checkpoint 75 (PPO, 1.224 s), then evolution strategies on the weights against 10 and then 20 non-official seeds (6-25), a fall costing the time limit, generation picked on held-out seeds. 0.932 s officially; finishes 38 of 40 fresh seeds (56-95), mean 0.973 s. No other entry's weights. Simulation only.
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> dadudu/microduck-bubu-hill
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:
claude/hugging-face-model-versions-s42dvp - commit:
08c85aeb6
- Downloads last month
- 23