Introduction

Abstract

This chapter presents the motivation and objectives for this book, and an overview of the work presented in the book. I begin by presenting the motivation for applying reinforcement learning (RL) to robots. Next, I present four specific challenges for applying RL to robotics problems. Then I describe a particular challenge of learning in few enough samples to be effective on domains with limited, expensive samples such as robots. I then provide a brief overview of the texplore algorithm introduced in this book and how it addresses these issues. Finally I present the contributions of this book and preview of each chapter of the book.