Extending the OpenAI Gym for robotics: a toolkit for reinforcement learning using ROS and Gazebo

TL;DR

This paper introduces an extension of OpenAI Gym tailored for robotics, integrating ROS and Gazebo to facilitate benchmarking and comparison of reinforcement learning algorithms in simulated robotic environments.

Contribution

It presents a new toolkit that combines OpenAI Gym with ROS and Gazebo, enabling standardized benchmarking of RL algorithms in robotics.

Findings

Q-Learning and Sarsa achieved comparable performance in simulated tasks.

The system allows consistent comparison of different RL algorithms in robotics.

Benchmarking results demonstrate the toolkit's effectiveness for robotics research.

Abstract

This paper presents an extension of the OpenAI Gym for robotics using the Robot Operating System (ROS) and the Gazebo simulator. The content discusses the software architecture proposed and the results obtained by using two Reinforcement Learning techniques: Q-Learning and Sarsa. Ultimately, the output of this work presents a benchmarking system for robotics that allows different techniques and algorithms to be compared using the same virtual conditions.