In planetary exploration, one of the greatest challenges faced by researchers is understanding how rovers

will navigate extraterrestrial environments, especially in rocky, unstable terrains and low-gravity systems.

Quadruped robots are studied for this reason, as they are better equipped to navigate such terrain much more effectively than wheeled rovers. Methods of virtual simulations using the Boston Dynamics Spot robot, an agile quadruped robot, can be used to test how robotic systems can function in alternate conditions. This project aims to create an open-source code package for the Spot robot’s simulation and navigation using the latest versions of ROS2 (Robot Operating System) and Ignition Gazebo, a newer and more advanced robotic simulation software, on rocky terrain that mimics a lunar and/or Martian surface. Basic reinforcement learning (RL) techniques using a preconfigured Proximal Policy Optimization (PPO) algorithm utilized the data collected from experiments conducted in a simulated uneven terrain to enhance the understanding of the robot’s navigation in similar extraterrestrial conditions. In future applications, applying this RL algorithm to a simulation of the Spot robot on an accurate lunar will enhance researchers’ understanding of quadruped navigation in complex environments