RL algorithms, imitation, and robotics

Picture it

Value methods score actions; policy methods adjust action probabilities; actor-critic methods combine both. Imitation learns from demonstrations. A robot adds a hard constraint: mistakes happen in the physical world, not just a simulator.

What to understand

• Q-learning, policy gradients, actor-critic, PPO
• Model-based planning, imitation, offline RL, sim-to-real

Follow the references

1. 1 · Course + videos

   UC Berkeley · CS285 Deep Reinforcement Learning ↗

   Use introductory lectures on behavior cloning, value functions, and policy gradients. Trace what data each method needs.