CXRLGym: ROS 2 Reinforcement Learning Environment

The CXRLGym module provides a ROS 2-integrated Gymnasium environment for reinforcement learning (RL) with multiple robots.

It acts as a bridge between the CLIPS-Executive and standard RL frameworks, allowing agents to perceive and act in a symbolic robot environment using familiar Gym interfaces.

Overview

CXRLGym extends gymnasium.Env to provide a ROS-driven reinforcement learning interface. It automatically constructs its observation and action spaces by querying ROS 2 services and actions.

Key Features

• Dynamic observation and action spaces generated from ROS 2 services
• Multi-robot management and coordination via reinforcement learning
• Compatible with Gym-based RL libraries (e.g., Stable Baselines3)
• ROS 2 action interfaces for reset, robot allocation, and action execution