This repository provides a customizable ROS2 environment for training multiple 2D drive cars using Deep Q-Network (DQN). Currently, agents learn collision-free navigation by avoiding obstacles, but the environment is designed to allow easy modification to train for various objectives in the future.