ICRA2001 Tutorial

Probabilistic Techniques for Mobile Robots


Wolfram Burgard
Department of Computer Science
University of Freiburg
Georges-Koehler-Allee Geb. 079
79110 Freiburg, Germany

Phone: +49 761 203 8026
Fax: +49 761 203 8007
Email: burgard@informatik.uni-freiburg.de
http://www.informatik.uni-freiburg.de/~burgard
Dieter Fox
Department of Computer Science & Engineering 
University of Washington 
Box 352350 
Seattle, WA 98195-2350

Phone: (206) 543-1695 
Fax: (206) 543-2969 
E-mail: fox@cs.washington.edu
http://www.cs.washington.edu/homes/fox/
Sebastian Thrun
School of Computer Science
Carnegie Mellon University
5000 Forbes Ave.
Pittsburgh, PA 15213-3891

Phone: (412) 268-8077
FAX: (412) 268-5576
thrun@cs.cmu.edu
http://www.cs.cmu.edu/~thrun/

 

Come and learn how to build robots like

Rhino

and

Minerva

 

Table of Contents

  1. Introduction
  2. Probabilistic Reasoning
  3. Probabilistic State estimation
  4. Probabilistic Decision making
  5. Conclusions
  6. Summary
  7. Successful applications of probabilistic robotics
  8. Open research issues

Abstract

One of the fundamental problems in mobile robotics is to deal with the inherent uncertainty of the robot's sensors. Whenever a robot is installed in a real and possibly populated environment, it has to use its sensors to perceive information about the world in order to successfully fulfill its tasks. As robots are currently moving into service fields, the need for algorithms that can cope with unpredictable and complex worlds is immense. In recent years, there has been a flurry of activities on probabilistic methods for robotics, which has led to a range of novel representations and algorithms for robot perception and control. For example, probabilistic methods have led to new, more scalable solutions mobile robot localization, exploration, mapping, and distributed robot control.
 
This  tutorial will provide an introduction to contemporary probabilistic techniques for sensor interpretation, state estimation and robot control. It will give a systematic overview of recent attempts to apply statistically sound algorithms and representations to robotic problems. Using examples such as mobile robot localization and mapping, we will demonstrate how the probabilistic approach yields different - and often more robust - solutions than previous approaches in robotics. The tutorial will also discuss in depth examples of specific robotic systems, to point out compromises one has to make to transfer statistical ideas into robot systems acting in the real and dynamic environments.
 
Since this tutorial discusses many innovative approaches from the field of mobile robotics and especially navigation and control, it might be highly attractive to students and researchers working in these areas. The methods we present can also be applied to other areas of robotics, so that it should also be interesting to people from other research fields within robotics. Furthermore, we present several robots which have been successfully deployed in real world applications. Thus, this tutorial might also be attractive to people from industry.