Reinforcement learning (RL) is an area of machine learning where an agent learns to make decisions itself by taking actions in the environment. This is useful in robotics because we could use RL to train our robot to do tasks (say walk) by itself, rather than programming it to do so. Generally, we train an RL agent in simulation, and then transfer the learned policy to the physical robot. Probably the best place to start learning RL is the classic Reinforcement Learning: An Introduction book by Sutton and Barto. I recommend going through the first three chapters to get a good understanding of the basics and terminology used in RL. After that, these posts might be helpful.

  1. The goal of Reinforcement Learning October 11, 2025