"Probabilistic Robotics" one: Preface

Catalog: Slightly

Objective

The purpose of this book is to introduce the new research field of "Probabilistic robotics". Probabilistic robotics is a branch area of robotics related to intelligence and control. It is characterized by the use of probabilistic and statistical methods to record information and action decisions. In the face of the unavoidable "uncertainty" problem in the use of robotics, we believe that robotics must aim at this approach. The algorithm design using probability theory has become a powerful framework in robotics in recent years. This book is the first academic book to introduce some of the main methods in this field.

This book places an important position on the description of the algorithm. All of the algorithms in the book are based on a fundamental theory: Bayes ' Law, and Bayesian filters that are extended to apply them to time series data. The probability-based algorithms presented in this book will have such a consistent mathematical background.

In writing, the book tries to describe the details of each method in the most complete form. Each chapter has one or more main algorithms introduced, the description of each algorithm is generally composed of the following 4 parts:
(1) sample implemented using pseudo-code
(2) Derivation of the formula starting from axiom. This part is to illustrate the various prerequisites behind the implementation of the algorithm is indispensable.
(3) A large number of experimental results. The main purpose of these experimental data is to help readers understand the algorithm in more detail.
(4) For each algorithm, from the standpoint of the people who implement the algorithm, discuss its good place and bad place in detail.
It is quite a laborious task to complete the description of the above parts for each algorithm. And the result of this is that for those readers who want to read quickly, much of the content becomes difficult to understand (of course, the formula deduction is probably the part that can be skipped). We also expect readers to be able to have a deep understanding of the field through this book.

This book is a collection of more than 10 years of research work with our students and many researchers in the field. We began writing this book in 1999, when we thought we could finish writing in just a few months. But after 5 years, the original draft basically did not say how much to stay, the content has changed. In the meantime, by writing this book and learning more about the theory of information and action decisions than I imagined, it is a very happy thing to be able to pass on many of the things we have learned through this book.

This book is an academic book, suitable for students, researchers, and other people who need to install robotics. We believe that all robot makers need software development, so we think the content of this book should be of interest to all robot researchers. At the same time, outside the robotics field, we think this book can also be of interest to researchers who apply statistics or who use sensors to observe the real world. For readers with different technical backgrounds, the book also carefully adjusted the narrative way, so that the reader can only rely on this book to understand the content. While reading linear algebra and basic probability statistics is helpful for reading this book, the basic probability rule is introduced and the most advanced mathematical knowledge is avoided as much as possible.

In addition, this book can also be used for handouts. Questions and participation exercises are set up in each chapter. When used as a handout, each chapter will have a better rhythm than one or two words. There is no need to follow the catalogue order when explaining, can be appropriate to interspersed or skipped some chapters. In fact, when we teach in this book, we usually start with a lecture in the seventh chapter or so. Use this book to learn, refer to the chapter at the end of the actual case, the practice of the way to complete the experiment is a more respected method. There is nothing more important than to implement robotics technology on your own.

We have gone all out, but it is still inevitable that the contents of the book or some technical imperfections. Accordingly, the relevant amendments are also communicated in the website of this book:
www.probabilistic-robotics.org
The site also has information about the book. I hope you will find something exciting in this book.

Sebastian Thrun
Wolfram Burgard
Dieter Fox

"Probabilistic Robotics" one: Preface