Welcome to the RL for Robotics Trainer

A structured browser course aligned with a reinforcement learning for robotics curriculum: Markov decision processes, value functions, tabular and deep algorithms, policy optimization in simulation, sim-to-real transfer, safety constraints, offline learning, multi-task policies, and an integrative capstone—with quizzes and scenarios.

4
Learning Levels
29
Topics
87
Exercises
29
Scenarios

What this trainer includes

  • probability, MDPs, Bellman equations, DP, Monte Carlo, TD, Q-learning and SARSA
  • function approximation, DQN, policy gradients, actor-critic, PPO, simulators, reward shaping, IL
  • model-based RL, domain randomization, SAC, TD3, HER, safe RL, locomotion and manipulation
  • offline RL, goal-conditioned and multi-task learning, robotics foundation models, capstone
  • English and German; local progress in the browser