Three-link 2D biped walker modeling and controlling with PDcontroller and reinforcement learning

Mini-project to learn a biped robot to walk using Deep Reinforcement Learning. Check LeggedRobotsProject.pdf for more information.

Requirement

Matlab 2019 or higher. Reinforcement Learning Toolbox.

To test

Launch the function set_path.m, load an agent (from the folder saved_agent) in the workspace and launch the function testAgent.m.

To train

Launch the function set_path.m, then the function trainAgent.m