Training robots to perform tasks with machine learning instead of hand-coded motions. Covers imitation learning from teleoperation demonstrations, reinforcement learning, vision-language-action models, sim-to-real transfer, and evaluating policies on hardware.