Gym-Gazebo-Sim brings RL agents to ROS2 Jazzy and Gazebo Harmonic

The open-source robotics community has a new tool for reinforcement learning: gym-gazebo-sim. Developed by fizzym and contributors, this toolkit is a direct successor to erlerobotics' popular gym-gazebo framework, but updated for ROS2 Jazzy and Gazebo Sim Harmonic. It allows users to train RL agents using the same familiar interface, now compatible with the latest ROS2 and Gazebo versions. A key design goal is low hardware requirements – the toolkit is engineered to run on low-performance computers, making it ideal for educational settings where students may not have access to powerful workstations. The developers are using gym-gazebo-sim in an introductory robotics undergraduate class, ensuring it gets real-world testing and continuous improvement.

For professionals and educators, this fills a critical gap. Most existing RL frameworks for robotics still rely on ROS1 and Gazebo Classic, which are being phased out. Gym-gazebo-sim provides a modern, maintained alternative that leverages the latest fixes and features of ROS2 Jazzy and Gazebo Harmonic. The toolkit supports standard RL training workflows, including environment setup, reward functions, and agent integration. By lowering the barrier to entry with low-performance computer support, it enables more institutions and hobbyists to experiment with reinforcement learning for robotics. The project is open-source and actively developed, with the community encouraged to contribute and adapt it for their own curricula or research projects.