Community-trained MicroDuck chimney-climb RL policy ships fully reproducible on Hugging Face
victormustar · x · 2026-10-01
HF community member HannesVonEssen released microduck-chimney-climb: three RL policies (climb/enter/exit) letting a MicroDuck robot brace between two walls to climb a narrow vertical corridor, reusing Pollen Robotics' official recovery policy. It's simulation-only, unvalidated on hardware, but ships full source, repro commands, and pinned checkpoints — and HF's Victor Mustar teases one-click deployment of community policies to real robots.
More from Embodied
- EngineAI T800 hailed as China's most dynamic robot, but fell more than any other at IROS — chris_j_paxton · 2026-10-01
- Custom Retro Hardware Is a Whole New World with AI for Hardware Design — pvncher · 2026-10-01
- Researcher dismantles Figure's IP excuse for destroying F.02: "awful marketing" — MarwaEldiwiny · 2026-10-01
- Autonomous's $20,900 Solar WorkPod Offers a Backyard Office with Dual RTX 5090 AI Datacenter — dee_hw · 2026-10-01
- Boston Dynamics' four-finger hand praised: balancing capability and simplicity — MarwaEldiwiny · 2026-10-01
- EPFL's TERRA pipeline teaches muscle-actuated agents to walk 9.4 hours of diverse terrain — amathislab · 2026-10-01