Using OpenAI with ROS Course - Python

Use the power of OpenAI combined with ROS simulations the easiest way.

Course Summary

In this Course, you are going to learn how to use the OpenAI ROS structure developed by The Construct and how to generate new code for it. The OpenAI ROS structure will allow you to develop for OpenAI with ROS in a much more easy way.

What you will learn

Basic Concepts of the OpenAI ROS structure
Set up the OpenAI ROS structure for a CartPole environment
Train the Cartpole with the qlearn algorithm
Set up the OpenAI ROS structure for a Moving Cube environment
Train the Cube with the qlearn algorithm
Modifying the learning algorithm: DeepQ
Set up the OpenAI ROS structure for a Fetch Robot
Training Fetch robot with the HER algorithm from OpenAI baselines

Course Overview

Introduction to the Course

Unit for previewing the contents of the Course.

Exploring the OpenAI Structure: CartPole

Follow, step by step, the full workflow of a CartPole simulated environment, including all the environments and scripts involved in its training.

Exploring the OpenAI Structure: RoboCube. Part 1

Learn how to apply the openai_ros package to your own robot.

Exploring the OpenAI Structure: RoboCube. Part 2

Learn how to create a Robot Environment for a Moving Cube with a single disk in the roll axis using the OpenAI ROS structure.

Exploring the OpenAI Structure: RoboCube. Part 3

Learn how to define the learning task of your robot by creating a Task Environment for a Moving Cube with a single disk in the roll axis. Also, you will use the Qlearn algorithm for training the RoboCube.

Save and Load the Learned Policy

Learn how to save the learned policy and how to load it to apply what the agent has learned.

Modifying the learning algorithm: CartPole

Learn how to set up the environment in order to be able to use the OpenAI Baselines deepq algorithm.

Modifying the learning algorithm: RoboCube

Learn how to set up the environment in order to be able to use the OpenAI Baselines deepq algorithm.

Training a Fetch Robot. Part 1

A step-by-step look at how to build the Robot Environment for training a Fetch robot.

Training a Fetch Robot. Part 2

A step-by-step look at how to build the Task Environment for training a Fetch robot.

Project: Training a Hopper robot

Create all the environments needed in order to be able to train the Hopper robot.

Teachers

Ricardo Tellez

Dreaming of a world where robots actually understand what they are doing. Developing the definitive tool that will make it happen.

Miguel Angel Rodriguez

Crashing engineering problems. Building solutions.

Alberto Ezquerro

Making easier the way the people learn how to program robots.

Robots used

CartPole Sim robot

Cube Sim robot

Fetch robot

Hopper robot

Learning Path

Machine Learning for Robots