gym-gazebo is an OpenAI Gym extension for Gazebo. This extension makes use of an ROS-Gazebo combination to train robots with reinforcement learning algorithms. In the previous chapters, we saw how ROS and Gazebo can be used to solve robotics applications or proof of concepts in simulations that, to an extent, emulate reality. This extension helps us to use this combination in controlling such robots in simulation. A simplified block diagram representation of gym-gazebo is shown as follows:

This is the underlying architecture of gym-gazebo. It consists of three main blocks—out of these, we already know Gazebo and ROS and their advantages. The third block makes use of the OpenAI Gym libraries and helps define environments and robots that are understood by ROS and Gazebo. The creators have created a set of custom environments in this block that can be understood by Gazebo. Alternatively, ROS can be used to control the robots in Gazebo through the plugins that we saw in Chapter 3, Building an Industrial Mobile Manipulator. Robots are defined as catkin workspaces that establish the Gazebo-ROS connection. Many worlds and environments have been created by this team. However, we're going to see only one robot with one environment in this book.

We will now look at TurtleBot and its environment.