31. Train robots through trial and reward
Use reinforcement learning for control and decision-making when trial, reward, and long-term outcome matter. Cover reward design, exploration, simulation training, safety constraints, and sim-to-real transfer.