A library containing Reinforcement Learning algorithms to train policies for quadruped robots.
This repository contains three main folders:
-
docker
Contains a Dockerfile and Docker Compose configuration files to build and run a Docker image with ROS2 Humble, IsaacSim, and IsaacLab already configured for use.
-
rl_quadrupeds
Contains several Python packages with custom functionalities for training Reinforcement Learning policies for quadruped robots. It also includes a tasks folder with the complete definitions of already trained RL policies.
-
ros2
Contains ROS2 Humble packages that enable interaction between RL simulations and ROS2 nodes.
Refer to each main folder's README file for setup instructions.