pollen-robotics/microduck_rl
RL training environments for Microduck (mjlab)
Build with Backblaze B2
SDKs, agent skills, IDE extensions, and reference pipelines from Backblaze Labs. All open source.
Loading star history...
Where it stands
Rank, growth pace and release cadence relative to everything GStars tracks.
The project lists 9 contributors on GitHub.
Peaked at #11 in the weekly ranking the week of Aug 30, its best in the past year.
Use Cases & Benefits
- Provides reinforcement learning training environments for the Microduck bipedal robot using MuJoCo Warp and PPO.
- Delivers high-fidelity sim2real transfer with detailed actuator physics, domain randomization, and backlash simulation for robust real-world deployment.
- Use for developing and training walking and balance policies for small bipedal robots in simulation before real-world testing.
- Use for experimenting with complex robot maneuvers like rolling, skating, and ball kicking using learned policies.
- Use for research on sim2real challenges and actuator modeling in robotics with a focus on voltage control and mechanical backlash effects.
Should you depend on it?
License terms, project hygiene and risks, from GitHub metadata and GStars' snapshots.
License
Apache 2.0
Permissive: commercial use, modification and redistribution are allowed with attribution; includes an explicit patent grant from contributors.
Community files
- README: present
Watch-outs
- No GitHub releases; pin a commit or tag instead of a release.
General guidance from repository metadata, not legal advice. How we score
From the README
RL training environments for Microduck — a ~800 g, ~25 cm tall bipedal robot — built on mjlab (MuJoCo Warp) with PPO. Policies are trained here at 50 Hz, exported to ONNX, and deployed on the real robot by the runtime in pollen-robotics/microduck.
The repo encodes the full sim2real recipe: BAM actuator physics, domain randomization, backlash simulation, and the reward-design lessons that made it work (see AGENTS.md for the distilled playbook).
Requires a CUDA GPU (training runs through MuJoCo Warp) and uv.
On ARM boxes (DGX Spark / GB10, Jetson):
uv syncpulls ~2 GB of CUDA wheels on first run and uv's default 30 s HTTP timeout can abort mid-download. ExportUV_HTTP_TIMEOUT=600for the first sync.No GPU? Add
--hf-jobsto any train command to run it on Hugging Face Jobs instead of locally (see scripts/hf/README.md).
uv run list-envsprints the live registry. Flat/Rough variants exist where noted.At deployment the runtime hot-swaps these policies (walk / recover / trick) behind a shared 61-dimensional observation contract, so any of them can take over the robot at any moment.
scripts/infer_policy.pyrehearses exactly that:Keyboard-driven (velocity commands,
Gground pick,Ysit/stand,Rroulade,K/Lkicks);--debug,--save-csv,--recordsupport sim2real comparisons.
Daily snapshots
Stars gained per day, and how forks and open issues moved, from GStars' daily snapshots
+2,081 stars over 68 days across 13 snapshot days; best day +363 on Aug 31.
Forks
679+547
Open issues
61+53
Compared with peers in Machine Learning
Fastest-growing repositories GStars tracks in this cohort for the week of Nov 1, by 7-day star growth
| Repository | Stars | +7d | Health | Last release |
|---|---|---|---|---|
| photoprism/photoprism | 40.4K | — | 65 | 1 day ago |
| huggingface/diffusers | 34.8K | — | 65 | 3 days ago |
| onnx/onnx | 21.6K | — | 63 | 31 days ago |
| lukas-blecher/LaTeX-OCR | 16.6K | — | 18 | — |
| milesial/Pytorch-UNet | 11.7K | — | 18 | — |
| pollen-robotics/microduck_rl | 2.8K | — | 12 | — |
| Category median | 21.6K | — | 63 | — |
Also by pollen-robotics
Other repositories from this maintainer that GStars tracks
Discover Repositories
Search across tracked repositories by name or description