# Gym Gazebo Sim - Reinforcement Learning using ROS2 Jazzy and Gazebo Harmonic

**URL:** <https://discourse.openrobotics.org/t/gym-gazebo-sim-reinforcement-learning-using-ros2-jazzy-and-gazebo-harmonic/56807>\
**Category:** Training & Education\
**Tags:** ros2, gazebo, deep-learning, jazzy, harmonic\
**Created:** [July 20, 2026, 8:07pm UTC](https://discourse.openrobotics.org/t/gym-gazebo-sim-reinforcement-learning-using-ros2-jazzy-and-gazebo-harmonic/56807 "2026-07-20T20:07:21Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![fizzym](https://avatars.discourse-cdn.com/v4/letter/f/ed655f/32.png) [@fizzym](https://discourse.openrobotics.org/u/fizzym)\
**Post date:** [July 20, 2026, 8:07pm UTC](https://discourse.openrobotics.org/t/gym-gazebo-sim-reinforcement-learning-using-ros2-jazzy-and-gazebo-harmonic/56807/1 "2026-07-20T20:07:21Z")

</div>

Hello everyone,

We very much appreciated the erlerobotics gym-gazebo framework for training RL agents in ROS1 and Gazebo Classic. And we wanted to be able to do the same in ROS2 Jazzy and Gazebo Sim Harmonic. So we followed erlerobotics lead and released gym-gazebo-sim:

> **[GitHub - ENPH353/gym-gazebo-sim: API extension for Gymnasium in Gazebo Harmonic](https://github.com/ENPH353/gym-gazebo-sim/tree/master)**
>
> master

We will use this for our introductory robotics undergraduate class so we will evolve it as we use it.

The toolkit is designed to run on “low performance computers” and we hope that some of you can make use of it too.