Pollen Robotics’ microduck_rl repository provides reinforcement-learning environments for the Microduck bipedal robot, built on mjlab and MuJoCo Warp with PPO. According to the vendor, policies are trained at 50 Hz, exported to ONNX and deployed on the robot by the Microduck runtime. Documented tasks include walking with velocity and head-pose commands, walking with fall recovery, standing up, sitting and standing, and other movements. The repository also documents sim-to-real techniques including BAM actuator physics, domain randomization and backlash simulation. The code is licensed under Apache License 2.0.