Book Image

Hands-On ROS for Robotics Programming

By : Bernardo Ronquillo Japón
Book Image

Hands-On ROS for Robotics Programming

By: Bernardo Ronquillo Japón

Overview of this book

Connecting a physical robot to a robot simulation using the Robot Operating System (ROS) infrastructure is one of the most common challenges faced by ROS engineers. With this book, you'll learn how to simulate a robot in a virtual environment and achieve desired behavior in equivalent real-world scenarios. This book starts with an introduction to GoPiGo3 and the sensors and actuators with which it is equipped. You'll then work with GoPiGo3's digital twin by creating a 3D model from scratch and running a simulation in ROS using Gazebo. Next, the book will show you how to use GoPiGo3 to build and run an autonomous mobile robot that is aware of its surroundings. Finally, you'll find out how a robot can learn tasks that have not been programmed in the code but are acquired by observing its environment. You'll even cover topics such as deep learning and reinforcement learning. By the end of this robot programming book, you'll be well-versed with the basics of building specific-purpose applications in robotics and developing highly intelligent autonomous robots from scratch.
Table of Contents (19 chapters)
1
Section 1: Physical Robot Assembly and Testing
5
Section 2: Robot Simulation with Gazebo
8
Section 3: Autonomous Navigation Using SLAM
13
Section 4: Adaptive Robot Behavior Using Machine Learning

Preparing the environment with TensorFlow, Keras, and Anaconda

Together with Anaconda, which you were instructed to install in the previous chapter, you will now install the machine learning tools TensorFlow and Keras. You will need them to make the neural networks that are required to solve the reinforcement learning tasks:

  • TensorFlow is the low-level layer of your machine learning environment. It deals with the mathematical operations involved in the creation of neural networks. Since they are mathematically resolved as matrix operations, you need a framework that is effective at solving this algebra, and TensorFlow is one of the most efficient frameworks for that. The name of the library comes from the mathematical concept of a tensor, which can be understood as a matrix with more than two dimensions.
  • Keras is the high-level layer of your machine learning environment. This...