Microduck RL
Summary
pollen-robotics/microduck_rl provides RL training environments for the Microduck robot using MuJoCo Warp and PPO. The project documents a full sim-to-real workflow, including BAM actuator models, domain randomization, and a 61-dimensional observation contract that enables runtime policy swapping. It includes detailed task definitions, training scripts, ONNX export, and deployment guidance, along with references to related tools and projects.