IJCAI-99 Workshop on

Reasoning with Uncertainty in Robot Navigation

On-line Proceedings

[ Workshop homepage ]


Mark Twain once said: ``All you need in this life is ignorance and confidence, then success is sure.'' Although there may be some truth in this, most scientists in robotics aim for robotic behaviour based on knowledge rather than on ignorance. However, in many cases, a robot has no perfect knowledge of its environment and of the results of its actions, and it needs to deal with uncertainty at many levels. In robot navigation, uncertainty typically plays a role in tasks like sensor interpretation, sensor fusion, map making, path planning, self-localisation, and control.

Dealing with uncertainty constitutes the focus of a large research effort in AI, which has led to the development of a number of new theories and new techniques. In the field of robot navigation no such research tradition exists, but recently several of the new approaches to uncertainty have been used to address the issues of uncertainty in robot navigation. Examples of these approaches are fuzzy set theory, possibility theory, Dempster-Shafer theory, decision-theoretic procedures like partially observable Markov decision processes (POMDPs), and Bayesian nets.

Recent successes of mobile robots in practical areas such as robotic museum tour-guides show that present-day navigation techniques, often incorporating both more or less traditional statistical methods and some of the mentioned new techniques, have reached a level of robustness which allows robots to operate even in crowded environments. Still, many problems, both practical and theoretical, remain. For example, the following questions raised during the international workshop on ``Reasoning with Uncertainty in Robotics'' (Amsterdam, NL, December 1995), which was probably the first to bring together researchers in the field of uncertainty in AI and robotics, are still open.

As mobile robots are going to be used for increasingly complex tasks, it is likely that the issues of dealing with uncertainty will keep cropping up. This warrants a continuing interest in the systematic study of these issues.

The present workshop is intended to discuss advanced methods of dealing with sensor and movement uncertainty in mobile robots. In particular, ideas about how to extend current theories to deal with more challenging environments will be a central topic. Hopefully, the workshop will provide an opportunity to critically examine various (competing) approaches, and to discuss their strengths and weaknesses in several types of environments.

The subject of the workshop should be of interest to all people working in mobile robotics. It is also expected to be of interest to researchers in the area of uncertainty in AI who welcome the possibility to apply their techniques to real and challenging problems.

Dieter Fox, Kurt Konolige, Alessandro Saffiotti, Frans Voorbraak
Organising Committee

Invited talks
[636K]  Statistical algorithms for mobile robot navigation.
Sebastian Thrun
Carnegie Mellon, USA

       POMDPs as a basis for map learning and navigation.
Leslie Kaelbling
Brown University, USA

Contributed papers

The following papers are available in PostScript format, compressed using gzip. File size is indicated on the left.

[249K]   Robust mobile robot localisation from sparse and noisy proximity readings.
A. Grossmann and R. Poli
Birmingham University, United Kingdom

[158K]   Active global localisation for a mobile robot using multiple hypothesis tracking.
P. Jensfelt and S. Kristensen
Royal Institute of Technology, Sweden, and DaimlerChrysler, Germany

[112K]   Hierarchical decision-theoretic planning for autonomous robotic surveillance.
N. Massios and F. Voorbraak
Amsterdam University, The Netherlands

[431K]   On-line selection of stable visual landmarks for a mobile robot under uncertainty.
I. Moon, J. Miura and Y. Shirai
Osaka University, Japan

[1,098K]  Utility theoretic planning in a behavior-based system.
J. Rosenblatt
University of Sydney, Australia

[1,217K]  Merging probability and possibility for robot localization.
C. Sossai, P. Bison, G. Chemello and N. Trainito
CNR Padova, Italy
Ramp session

In order to encourage active participation by everyone, the attendants who did not present a full-length paper have been given the opportunity to give a short outline of their research in an informal ``ramp session.'' Possible oulines range from summary of a recent relevant paper that the author has published elsewhere, to a position statement, to the report of current, fresh, half baked work. The oulines are accessible by clicking on the numbers on the left.

[ 1.E. Araujo and R. Grupen (University of Massachussetts, MA, USA)

[ 2.]  H. Asoh (Electrotechnical Lab., Japan)

[ 3. P. Fabiani and Y. Meiller ( Onera Toulouse, France)

[ 4.]  A. Howard (University of Melbourne, Australia)

[ 5.]  K. Konolige and K. Chou (SRI International, CA, USA)

[ 6.]  K. Murphy (University of Berkeley, CA, USA)

[ 7.]  M. Oskarsson and K. Åström (Lund University, Sweden)

[ 8.L. Sucar and E. Morales (ITESM, Mexico)

Last update: Aug 30, 1999
Maintainer: A. Saffiotti
Hosted by:  AASS Home