53. Reinforcement learning for robotics

Model robot behavior as states, actions, rewards, and policies. Apply value-based and policy-gradient methods while addressing sample cost, unsafe exploration, sparse rewards, and reproducibility.