Reinforcement Learning Accelerates Humanoid Behavior Production

Reinforcement learning is used to speed the production of behavior for the Atlas humanoid robot. At the heart of the learning process is a physics-based simulator that generates training data for a variety of maneuvers. The control policy tracks and controls retargeted human motion data. Each maneuver is created with data from about 150 million runs of the simulator and transferred zero-shot to the hardware. This work is part of a collaboration between the RAI Institute and Boston Dynamics.