REASONING WITH UNCERTAINTY IN ROBOTICS International workshop University of Amsterdam, The Netherlands December 4-6, 1995. AIM: Interaction between research on uncertainty in AI and robotics. BACKGROUND In the field of artificial intelligence, a lot of effort is spent on the problem how to deal with information which is uncertain, incomplete, inconsistent, or otherwise deficient. This research has led to a large number of new theories and new techniques. In particular, in the area of numeric uncertainty, one can mention the certainty factor model, Dempster-Shafer theory, interval probability theory, probabilistic networks, probability logic, possibility theory, et cetera. In robotics, uncertainty plays an important role at many levels, such as image interpretation, sensor fusion, navigation, path planning, and control. Therefore, this field seems to be particularly well suited to test and apply the new theories and techniques mentioned above. However, uncertainty in robotics is typically approached by more or less traditional statistical methods. On the other hand, research on uncertainty in AI is rather focused on theoretical work or on applications in the area of diagnosis. The workshop is intended to bring together researchers in the fields of uncertainty in AI and robotics in order to clarify the relation between the two research fields, to find out what the research in the two fields can mean for each other, and to identify the most urgent problems that have to be attacked. PROGRAMME The programme of the workshop includes tutorial lectures, invited talks, the presentation of accepted papers, and a forum discussion. 1. TUTORIAL LECTURES o tutorial on robotics (Crowley) o tutorial on uncertainty in AI (Voorbraak) 2. INVITED TALKS o navigation (Elfes) o sensor fusion (Durrant-Whyte) o map learning (Kaelbling) o image interpretation (Hummel) 3. PRESENTATION OF ACCEPTED PAPERS Papers are welcome in the whole field of reasoning with uncertainty in robotics, but papers discussing (either abstractly or by means of a concrete application) the relevance for robotics of the above mentioned AI approaches to uncertainty are preferred. In the papers, special attention has to be given to the informal underlying motivations and to the possible general conclusions that can be drawn from the work reported on. (How essential is the choice for some particular formalism? What are the conditions under which the formalism is applicable? On which points is there still room for improvement? Et cetera.) Possible topics include o sensor fusion - How to combine several (possibly conflicting) readings of a sensor? - How to aggregate evidence obtained from different types of sensors? o interpretation of uncertain geometric data - How to obtain 3D models from 2D images? - How to reduce or eliminate uncertainty in sensor data? o map learning - How to learn the spatial layout of the environment? - How to extract the relevant spatial structure? o navigation and path planning under uncertainty - How to navigate, given a map of the environment, sensor data, and the uncertainty in control? - How to navigate with limited prior knowledge of the environment? o decision under uncertainty - How to decide between action and the reduction of uncertainty? - How to formulate a decision theory for resource-bounded agents? 4. FORUM DISCUSSION "Towards the formulation of benchmark problems in the field of uncertainty in robotics" Ideally, the forum discussion will result in the formulation of one or two benchmark problems that can (and will) be used to compare different approaches to uncertainty in robotics and to measure progess in the field. Some relevant questions are - What are the bottlenecks for handling uncertainty in robotics? - What are the most urgent problems in the field? - Which of these problems can be attacked today or in the near future? - How can these problems be formulated as benchmark problems? SUBMISSION REQUIREMENTS Authors are asked to submit four (4) copies of their papers to the submission address given below by July 1st, 1995. Papers should have a front page containing: title, name, full address, email address and fax # (if available) for all authors, a list of keywords identifying the subject area of the paper, and a 100-200 word abstract. Papers should be a maximum of 15 pages (excluding front page) and printed on A4 paper in 12 point type with a maximum of 38 lines per page and 75 characters per line (corresponding to LaTeX article style, 12 pt). Submission of hard copies is strongly preferred. Electronic submission is acceptable, provided standard LaTeX is used. Notification of acceptance/rejection will be mailed to the first (or designated) author of each paper by September 1st, 1995. Accepted papers will be included in the informal proceedings distributed at the workshop. Accepted papers must be presented at the workshop, in English, by one of the authors. A commercial publisher is contacted to publish the final versions of the presented papers, which are due at February 1st, 1996, so they can benefit from the interaction at the workshop. ATTENDANCE The workshop will be limited to at most hundred participants. The registration fee is Dfl 400 for early registration (before November 1st, 1995) and Dfl 500 for late registration. The registration fee includes lunches, refreshments, conference dinner, and workshop proceedings. A limited budget will be available for the support of participants which are faced with financial problems such as serious difficulties with exchanging currency. The workshop will take place in Amsterdam, close to the city center. Detailed registration information will be available no later than July 1995. One can consult the WWW. If you are interested in attending, please send an email to rur@fwi.uva.nl, then you will be kept informed automatically. TIMETABLE - submission deadline: July 1st, 1995 - notification of acceptance/rejection: September 1st, 1995 - deadline for early registration: November 1st, 1995 - workshop: December 4-6, 1995 - final version of papers: February 1st, 1996 SUBMISSION ADDRESS Michiel van Lambalgen Department of Mathematics and Computer Science, University of Amsterdam Plantage Muidergracht 24, 1018 TV Amsterdam, The Netherlands Tel: +31 20 525 6060, Fax: +31 20 525 5206, Email: rur@fwi.uva.nl PROGRAMME COMMITTEE James L. Crowley (IMAG-LIFIA, Grenoble, France) Leo Dorst (University of Amsterdam, The Netherlands) Hugh Durrant-Whyte (University of Sydney, Australia.) Alberto Elfes (Automation and Software Systems Institute, Campinas, Brazil) Robert Hummel (Courant Institute of Mathematical Sciences, New York, USA) Leslie Pack Kaelbling (Brown University, Providence, USA) Michiel van Lambalgen (University of Amsterdam, The Netherlands) Frans Voorbraak (University of Amsterdam, The Netherlands) PROGRAMME CHAIR Frans Groen (University of Amsterdam, The Netherlands) ORGANIZING COMMITTEE Leo Dorst, Michiel van Lambalgen, and Frans Voorbraak (University of Amsterdam, The Netherlands) LOCAL ARRANGEMENTS Marco de Vries Plantage Muidergracht 24, 1018 TV Amsterdam, The Netherlands Tel: +31 20 525 6051, Fax: +31 20 525 5206, Email: rur@fwi.uva.nl SPONSORED BY PIONIER project Reasoning with Uncertainty, of the Netherlands Organization for Scientific Research (NWO) Institute for Logic, Language, and Computation (ILLC)