Program & Exhibit Guide

EIGHTEENTH NATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE (AAAI-02)

FOURTEENTH CONFERENCE ON INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE (IAAI-02)

July 28-August 1, 2002

Shaw Conference Centre and Westin Edmonton Hotel Edmonton, Alberta, Canada

SPONSORED BY THE AMERICAN ASSOCIATION FOR ARTIFICIAL INTELLIGENCE

Cosponsored by ACM/SIGART, Alberta Informatics Circle of Research Excellence (iCORE), Defense Advanced Research Projects Agency (DARPA), NASA Ames Research Center, NSF's Directorate for Computer and Information Science and Engineering (CISE), Naval Research Laboratory

In cooperation with The University of Alberta Contents Robot Challenge Cochairs Ben Kuipers, University of Texas at Austin and Acknowledgments / 2 Ashley Stroupe, Carnegie Mellon University Awards / 3 Conference at a Glance / 5 Robot Rescue Competition Cochairs DC-02 / 4 Jenn Casper, Mark Micire, and Robin Murphy, Exhibition / 18 University of South Florida General Information / 33 IAAI-02 Program / 13–15 Robot Host Competition Cochairs Intelligent Systems Demonstrations / 21 David Gustafson, Kansas State University and Invited Talks / 10 Francois Michaud, Universite de Sherbrooke Maps / 37–39 Robot Exhibition Cochairs National Botball Exhibition / 31 Ian Horswill and Christopher Dac Le, Poster Sessions / 3 Northwestern University Registration / 32 Robot Building Laboratory / 7 Mobile Robot Workshop Chair Robot Competition and Exhibition / 24 Bill Smart, Washington University St. Louis Special Events and Programs / 3–4 Special Meetings / 4 Robot Building Laboratory Chair Sponsoring Organizations / 2 David Miller, KISS Institute for Technical Program (AAAI / IAAI) / 37–39 Practical Robotics Tutorial Forum /6 Doctoral Consortium Chair Workshop Program / 8 Marie desJardins, University of Maryland, Baltimore County Acknowledgments Student Abstract and Poster Cochairs The American Association for Artificial Intel- Mark Craven, University of Wisconsin, ligence wishes to acknowledge and thank the Madison and Sven Koenig, Georgia Institute following individuals for their generous con- of Technology tributions of time and energy to the successful creation and planning of the Eighteenth Na- Tutorial Chair tional Conference on Artificial Intelligence Michael L. Littman, Rutgers University and the Fourteenth Conference on Innovative Workshop Chair and Cochair Applications of Artificial Intelligence. Berthe Y. Choueiry, University of Nebraska- AAAI Conference Committee Chair Lincoln and Janyce Wiebe, University of Pitts- James A. Hendler, University of Maryland burgh AAAI-02 Program Cochairs A complete listing of the AAAI-02 and IAAI- Rina Dechter, University of California, Irvine, 02 Program Committee members appears in Michael Kearns, University of Pennsylvania, the conference proceedings. and Richard S. Sutton IAAI-02 Conference Chair Sponsoring Organizations

Contents / Acknowledgments Acknowledgments / Contents Steve Chien, Jet Propulsion Laboratory AAAI gratefully acknowledges the generous IAAI-02 Conference Cochair contributions of the following organizations John Riedl, University of Minnesota to AAAI-02: ■ ACM/SIGART Intelligent Systems Demos Chair ■ Alberta Informatics Circle of Research Ex- George Ferguson, University of Rochester cellence (iCORE) Local Arrangements Chair ■ Defense Advanced Research Projects Agen- cy (DARPA) Russell Greiner, University of Alberta ■ NASA Ames Research Center Mobile Robot Competition and ■ NSF’s Directorate for Computer and Infor- Exhibition General Cochairs mation Science and Engineering (CISE) ■ Holly Yanco, University of Massachusetts Low- Naval Research Laboratory ■ ell and Tucker Balch, Georgia Institute of Tech- University of Alberta nology 2 Special Events and Programs 3 and , lowed in Ex- in Ex- PM ,fol AM AM in the Ball- PM PM invitation only). invitation

By .( PM ailable for questions. M. Mitchell, of Professor Fredkin Com- hnical Paper Poster Sessions Poster hnical Paper n conjunction with the AI Festival. with the n conjunction ad- In c esidentialAddress dnesday, 31, July 7:00 – 10:00 en and Kazuhisa Makino,en and Kazuhisa Osaka University dinner at 8:00 ,i ll give the AAAI Presidential Address on “AI on Address AAAI Presidential the ll give

nference. will be available. Refreshments ogram Cochairs Rina Dechter and Michael and ogram Rina Dechter Cochairs om on the second floor ofom on the second Ed- Westin the ew process, members of Com- the Program earns will the authors with present their cer- hosen to avoid conflicts of conflicts avoid hosen to interest, reviewed ouraged to attend this integral portion attend ouraged to of the hibit Hall C ofhibit Hall Centre. Conference the Shaw hibit Hall C ofhibit Hall Centre. Conference the Shaw Student Abstract Student Abstract Posters for in- chosen Students whose abstracts were will dis- proceedings clusion in the conference Poster Abstract at the Student their work play AB, Session in Exhibit Hall Conference Shaw 3:00 – 5:30 31 from Wednesday, July on Centre PM puter Science at Carnegie Mellon University, at CarnegieScience Mellon puter wi Te Sessions will be Poster Technical AAAI-02 The held on Tuesday, 30, July 5:45 – 8:45 by Pr Tom Brain in Sci- Revolution and the Impending ence” Wednesday, on 31, July 8:30 Outstanding Paper Award Paper Outstanding This year,Ar- on Conference National AAAI’s that ex- honors one paper tificial Intelligence contri- emplifies high in technical standards bution and exposition. During the blind re- vi consid- papers to which recommended mittee Award. Paper er for the Outstanding A subset of Committee,Senior Program the carefully c the winning papers and selected pa- all such per: Explana- All Abductive “On Computing tions” Universität Thomas Eiter, by Technische Wi Pr K on Tuesday,tificates 30, July at 8:00 Drawing Room of Drawing Room Macdon- Hotel the Fairmont ald.will begin reception at 7:30 A co ro monton. will papers Sixty technical be present- evening,ed each and will accord- be organized in (maps will area subject be available ing to the ballroom).AAAI-02 registrants en- are All c We dition,AAAI/SIGART participants in the ConsortiumDoctoral their poster will display during this session.presentations All students av be will in the PM Imperial College of iversity of Rochester iversity of California, ft Research Un o iversity of Pittsburgh Un Texas Tech University Un -Ilan University and Uni- -Ilan University r Carnegie University Mellon Micros Ba y E. Kyburg, Jr., in Exhibit Hall C of Hall in Exhibit Confer- the Shaw ael Gelfond, ael I. Jordan, ition g Award g Award ia P. Sycara, hell, President, AAAI and Bruce AM n to the author of n to pa- the most influential hanan,Awards and President AAAI Past llows Recognition Dinner Recognition llows versity of Maryland Kat Sarit Kraus, H.Stephen Muggleton, and UWF/Institute for Human & Machine for Human and UWF/Institute Cogn Mich Berkeley Science, and Medicine Technology Mich Eric Horvitz, Henr Kevin D. Ashley, tc mmittee Chair,mmittee Wednesday, on 31, July at c tificial Intelligence,Wash- in 1983 in held ll be given in the following three categories: in the following three ll be given ntributions the field of to artificial intelli- ve matic review, with non-trivial content, but the ward was established in 2000 to honor the au- was established in 2000 to ward ward is primarilyward for the exposition. Co 8:15 The 2002 Fellows Recognition Dinner will be Recognition The 2002 Fellows 30, July held Tuesday, 7:30 - 10:00 from ■ ■ ■ ■ ■ ■ ■ ■ gence, distinc- attained unusual and who have tion in the profession. an- AAAI is pleased to for 2002: the eight Fellows nounce newly elected Fe Artifi- for Association American the year Each of number a small recognizes cial Intelligence made significant sustained members who have co a The AAAI Effective The AAAI ExpositoryEffective Writin Writing Expository AAAI Effective The 2002 A thor(s) of a high quality, of piece effective writing, a broad the general public or to to accessible (not just a subarea),AI audience written within years.the last two The contribution should be based on sound science, ideas or sys- interesting te The AAAI Distinguished Service Award recog- Distinguished ServiceAAAI Award The for extraordi- one individual year each nizes naryAI community. service the to Distinguished ServiceAward The 2002 AAAI Classic Paper Award will be Award AAAI Classic Paper The 2002 gi on Conference National the Third per(s) from Ar ington, DC. Classic Paper Award Classic Paper The AAAI Awards will be presented by Tom by will be presented Awards AAAI The Mi Bu AAAI Recognition Award Recognition AAAI ence Centre.ence will in be announced Recipients the AAAI-02 Addendum.Program The awards wi Opening Reception tablished researchers. The thirteen students accepted to participate in this program will al- The AAAI-02 Opening Reception will be held so participate in the Student Poster program Monday, July 29, 7:00 – 10:00 PM,at Fort Ed- on Wednesday, July 31, from 3:00 – 5:30 PM monton Park (Canada’s largest living history during the AI Festival. All interested AAAI-02 park). This event will provide the traditional student registrants are invited to observe the opportunity for attendees to socialize in a presentations and participate in discussions at unique setting prior to the beginning of the the workshop. AAAI and ACM/SIGART grate- first day of technical sessions. A variety of hors fully acknowledge a grant from NSF’s Direc- d’oeuvres and a no-host bar will be available. torate for Computer and Information Science Admittance to the reception is free to AAAI-02 and Engineering (CISE) that partially sup- registrants including transportation to and ports student travel to the event. from the event. A $40.00 CDN per person fee ($15.00 CDN for children) will be charged for spouses and other nontechnical conference AAAI Business Meeting registrants. The AAAI Annual Business Meeting will be Shuttles will pick up attendees from the held Thursday, August 1, from 1:00 – 1:30 PM Shaw Conference Centre at the south side en- in Salon 6, meeting level, Shaw Conference trance by the upper level balcony. Shuttles will Centre. leave the Conference Centre beginning at 6:30 PM and continue until 7:30 PM.Return from Conference Committee Meeting Fort Edmonton will begin at 8:30 PM and the last shuttle will leave at 10:00 PM.Shuttles will The Conference Committee Meeting will be depart from Fort Edmonton at the main en- held Thursday, August 1, from 7:30 – 8:30 AM trance where attendees were dropped off, and in Salon 1, meeting level, Shaw Conference will return them to the conference hotels. Centre.

AI Festival AI Journal Editorial Board The AI Festival will be held in Exhibit Hall AB Meeting of the Shaw Conference Centre, Wednesday, The AI Journal Editorial Board lunch will be July 31 from 3:00 – 5:30 PM.This event will held Tuesday, July 30 from 12:30 - 2:00 PM in provide attendees the opportunity to stroll Salon 1, Meeting Level, Shaw Conference Cen- among numerous exciting events — the Mo- tre. bile Robot Competition and Exhibition, the Intelligent Systems Demos, the IAAI-02 En- trepreneurs Forum, the Student Posters, and a Executive Council Meeting competition between high school Botball The AAAI Executive Council Meeting will be competitors and the Robot Building Labora- AM PM ions,DC-02,and Meetings held Sunday, July 28, from 9:00 – 5:00 in tory participants — all enlivened by refresh- Turner Valley,second ﬂoor, The Westin Ed- ments and conversation. Admittance to the monton. Continental breakfast will be avail- festival is free to AAAI-02 registrants. A $25.00 able at 8:30 AM. pt CDN per person fee ($8.00 CDN for children) will be charged for spouses and other non- Program Committee Luncheon technical conference registrants. Guest tickets ce are available in onsite registration. The AAAI-02/IAAI-02 Program Committee Luncheon will be held Tuesday, July 30, 12:30 Doctoral Consortium (DC-02) – 2:00 PM in the River View Room, Pedway Level, Shaw Conference Centre to honor the Re The Seventh AAAI/SIGART Doctoral Consor- contributions of all the members of the AAAI- tium program will be held on Sunday and 02 and IAAI-02 Program Committees. (By in- Monday, July 28-29 from 8:30am – 5:30 PM in vitation only). the Saskatchewan Room, second ﬂoor of The Westin Edmonton. The Doctoral Consortium provides an opportunity for a group of Ph.D. students to discuss and explore their research interests and career objectives in an interdisci- plinary workshop together with a panel of es- 4 Conference at a Glance ession 5 istration Lab Building Robot g 31 30 29 ULY 28 ,J ULY ,August 1 ,August ULY kshops Workshops kshops Workshops ial Forum Forum Tutorial ial Forum Forum Tutorial r r ULY istration AAAI-02 / IAAI-02 istration Registration Session AAAI–02 Poster istration RegistrationS AAAI–02 Poster r r ,J ,J g g g MORNING AFTERNOON EVENING ,J to to Wo Wo ynote Address ynote AAAI-02 / IAAI-02 Dinner AAAI Fellows ot Building Labot Building Lab Building Robot Lab ot Building bot Workshop AAAI Business Meeting Re Re Re Registration Registration Opening Reception Registration Re Tu Tu esidential Addressesidential AAAI-02 / IAAI-02 Ke Ro ot Programs / Botballot Programs Festival AI ot Programs / Botballot Programs / Botball Programs Robot AAAI/SIGART DCAAAI/SIGART DC AAAI/SIGART AAAI/SIGART DC AAAI/SIGART DC AAAI-02 / IAAI-02 Exhibition / IS Demos AAAI-02 / IAAI-02Workshop Robot AAAI-02 / IAAI-02 Lunch Program Committee Rob Rob Pr EDNESDAY ONDAY Exhibition / IS Demos Programs /Botball Robot Exhibition / IS Demos Exhibition/IS Demos HURSDAY UESDAY UNDAY Rob Rob T W T M S Tutorial Forum CDN persyllabusinonsite registration. te At- Tutorial attendees may redeem theirtutorial syllabitickets atthetutorial rooms. schedules. fourconsecutive tutorials may be taken dueto parallel Amaximum of four tutorial syllabi. includesadmissiontoAAAI-02 technical upto registration fourtutorials andthecorresponding Forum Tutorial 6 for Combinatorial Algorithms Auctions andExchanges MA3: AITechniques for Personalized Recommendation (Full Day)MA2: onthe Integration Information Web MA1: 9:00 July 29 Monday, Session III: Machine for Practical Software Learning Engineering SP3: Collaborative Multi-Agent Systems SP2: Practical Approaches to Handling Uncertainty inPlanningandScheduling SP1: 2:00 -6:00 July 28 Sunday, Session II: Qualitative andTemporal Spatial Reasoning Aspects of SA4: Greedy On-LinePlanning SA3: Unique (Full Day) ChallengesandOpportunities AIinSpace: SA2: the inLanguageModeling Art TheState of SA1: 9:00 July 28 Sunday, Session I: ndees who wish to obtainsyllabifromndees whowish othe ao ,Shaw Conference Centre Salon 8, Tu Shaw Conference Centre Salon 10, A Shaw Conference Centre Salon 12, Cr Shaw Conference Centre Salon 9, Ti Shaw Conference Centre Salon 8, GroszBarbara andCharlieOrtiz Shaw Conference Centre Salon 10, J. Shaw Conference Centre Salon 9, Guesgen andHans W. Anger, Frank D. Gerard Ligozat, Shaw Conference Centre Salon 10, Sv Shaw Conference Centre Salon 12, Daniel Clancy Shaw Conference Centre Salon 8, Jo AM AM toyJmsn Joseph Konstan andJohn Riedl nthony Jameson, Christopher Beck and Thierry Vidal BeckandThierry Christopher shua GoodmanandEugene Charniak eze n ayD Boetticher m Menzies D. andGary en Koenig and Anthony Stentz aig Knoblock andSubbarao Kambhampati omas Sandholm - 1:00 - 1:00 PM PM PM r tutorials may purchase themseparately for$30.00 Tutorial Forum & Robot Building Lab 7 (RBL-02 Contest/Exhibition) (RBL-02 PM PM PM – 9:00 – 1:00 – 5:30 AM AM PM PM ael Wooldridge and Simon Parsons and Simon Wooldridge ael nday, 29 July 9:00 dnesday, 31 July 3:00 ch d Hogg, Carla P. Gomes, Walsh, Zhang and Weixiong Toby unday, 28 July 9:00 We Mo S Mi Salon 12, Centre Conference Shaw Ta Salon 8, Centre Conference Shaw be having a better time at the conference than logic the questions that can theorists? These are time at the conference a better be having eregistration is required. ex- to AI researchers for Lab (RBL) is a chance Building The Robot bot exhibition, afternoon.Wednesday on AI Festival during the is aimed at educa- The RBL

rs, in robotics. interested students and researchers of A general knowledge programming will quired) to bring to a MacOS,quired) 98, Windows with will or LINUX laptop that there them so be mul- Robot Building Lab Laboratory Building willThe Robot be held in Salon 2, meeting level, Centre Conference Shaw at the following times: MP2: Agents Rational Action in Autonomous 2:00 - 6:00 MP1: and Structure Problems in Combinatorial Transitions Phase Session IV:Session Monday, 29 July Pr periment with hardware.AI algorithm it actually gets em- when favorite your happens to What seem always do roboticists Why a simulation? to compared world the real is reliable bodied? How to participating by in the RBL.best be answered in the past, As RBL will the par- break this year’s small groups.ticipants into kit and then will day spend the next a robot will group Each be given and a half task. this year’s achieve to system a robot creating The lab will with conclude a friendly competition groups. among the different co- ‘multi-agent lab will be this year’s The theme for operation’. independent robots or more two kit will enough parts robot contain create Each to the task. accomplish (hopefully) to that will together work not (but encouraged are Participants re group.tiple programming stations for each of The results as part the lab will of be presented the ro to be assumed. is required.experience prior robotics No and taught The lab is being organized by AAAI. (KIPR) for Robotics for Practical the KISS Institute from and assistants are Instructors trained staff.KIPR’s is the lead instructor. Miller David Workshop Program

Attendance at the workshops is limited, and participation is by invitation only. All workshop participants must register for the AAAI-02 technical program. Registration onsite for a workshop is possible with the prior permission of the corresponding workshop organizer. The times for each workshop are listed below. Workshops will be held in the Shaw Conference Centre and The West- in Edmonton.

Sunday, July 28 W1: Agent-Based B2B Electronic Commerce Technologies Organizer: Brian Blake Consulate Room, The Westin Edmonton 8:30 AM – 6:00 PM W5: Autonomy, Delegation, and Control: From Inter-Agent to Groups Organizers: Henry Hexmoor and Rino Falcone Chairman Room, The Westin Edmonton 8:45 AM – 6:10 PM W7: Cognitive Robotics (CogRob2002) Organizers: Chitta Baral and Sheila McIlraith Salon 11, Shaw Conference Centre 9:00 AM – 6:00 PM W8: Game Theoretic and Decision Theoretic Agents Organizers: Piotr Gmytrasiewicz and Simon Parsons Salon 3, Shaw Conference Centre 9:00 AM – 5:30 PM

Workshop Program W9: Intelligent Integration of Information and Services on the Web Organizers: Dean Allemang, Eleni Stroulia and John Mylopoulos Salon 17/18, Shaw Conference Centre 9:00 AM – 4:30 PM W11: Meaning Negotiation Organizer: Paolo Bouquet Salon 15/16, Shaw Conference Centre 8:30 AM – 6:00 PM W15: Planning with and for Multiagent Systems (two-day) Organizers: Michael Brenner and Marie desJardins BC Room, The Westin Edmonton 2:00 – 6:00 PM W16: Preferences in AI and CP: Symbolic Approaches (two-day) Organizer: Ulrich Junker Salon 4, Shaw Conference Centre 9:00 AM – 5:15 PM W17: Probabilistic Approaches in Search Toby Walsh and Carla Gomes Salon 6, Shaw Conference Centre 9:00 AM – 6:00 PM W19: Semantic Web Meets Language Resources Organizers: Nancy Ide and Chris Welty Chancellor Room, The Westin Edmonton 9:00 AM – 6:00 PM

8 Monday, July 29 Program Workshop W3: Artiﬁcial Intelligence for Intelligent Business Organizers: Daniel E. O’Leary and Alun Preece Chancellor Room, The Westin Edmonton 8:30 am - 3:00 pm W4: Automation as Caregiver: The Role of Intelligent Technology in Elder Care (W4) Organizer: Karen Haigh Salon 11, Shaw Conference Centre 8:30 AM – 6:00 PM W6: Coalition Formation in Dynamic Multiagent Environments Organizers: Leen-Kiat Soh and Charlie Ortiz Consulate Room, The Westin Edmonton 8:45 AM – 4:30 PM W10: Intelligent Situation-Aware Media and Presentations (ISAMP) Organizer: Rainer Malaka and Antonio Krueger Salon 17/18, Shaw Conference Centre 9:30 AM – 6:00 PM W13: Multi-Agent Modeling and Simulation of Economic Systems Organizers: Koichi Kurumatani, Shu-Heng Chen and Azuma Ohuchi Salon 6, Shaw Conference Centre 8:55 AM – 5:00 PM W14: Ontologies for the Semantic Web Organizers: Adam Pease, Richard Fikes, and Jim Hendler Salon 15/16, Shaw Conference Centre 9:00 AM – 5:30 PM W15: Planning with and for Multiagent Systems (two-day) Organizers: Michael Brenner and Marie desJardins BC Room, The Westin Edmonton 9:00 AM – 6:00 PM W16: Preferences in AI and CP: Symbolic Approaches (two-day) Organizer: Ulrich Junker Salon 4, Shaw Conference Centre 9:00 AM – 1:00 PM W18: Real-Time Decision Support and Diagnosis Systems (Joint Workshop with KDD-02 and UAI hosted by AAAI-02) Organizers: Haipeng Guo, Eric Horvitz, William H. Hsu, and Eugene Santos Jr. Salon 3, Shaw Conference Centre 8:30 AM – 6:00 PM W20: Spatial and Temporal Reasoning Organizers: Hans W. Guesgen, Frank D. Anger, and Gerard Ligozat Chairman Room, The Westin Edmonton 9:00 AM – 5:00 PM

Thursday, August 1 W12: Mobile Robotic Competition and Exhibition Workshop Organizer: Bill Smart Salon 3, Shaw Conference Centre 9:00 AM – 5:00 PM

9 All AAAI-02 invited presentations will be held co-evolved to ensure the information extract- in Exhibit Hall C, assembly level, Shaw Con- ed by the robots is truly usable by the rescue ference Centre. IAAI-02 invited presentations community. will be held in Salon 8, meeting level, Shaw Extensive video footage of the site and Conference Centre. “robot’s eye” views will be shown.

Tuesday, July 30 Wednesday, July 31

8:30 – 9:30 AM 8:30 – 9:30 AM AAAI Keynote Address AAAI Presidential Address Probabilistic AI and AI and the Impending Revolution in Information Retrieval Brain Science Michael I. Jordan, Department of Electrical Engi- Tom M. Mitchell, Carnegie Mellon University neering and Computer Science & Department of Exhibit Hall C, Assembly Level, Shaw Conference Centre Statistics, University of California, Berkeley Tom M. Mitchell is the Fredkin Professor of Exhibit Hall C, Assembly Level, Shaw Conference Centre Computer Science at Carnegie Mellon Univer- Much progress has been made in recent years sity. He is author of the textbook Machine in the area of information retrieval, in partic- Learning, a member of the National Research ular as embodied in Internet search engine Council’s Computer Science and Telecommu- Invited Talks technology. Much progress has also been made nications Board, and President of the AAAI. in probabilistic, graph-theoretic AI. What are Mitchell’s research over the years has dealt the possibilities for bringing these two lines of with theoretical and practical issues in ma- research together — for viewing large-scale in- chine learning. During 2000-2001 he served as formation retrieval as a core enabling technol- Chief Scientist at WhizBang! Labs, a company ogy for AI systems, and for asking IR systems that employs machine learning to extract de- to exhibit true inferential capabilities. Jordan tailed factual information from text. Since re- will discuss research aimed at bridging the turning to Carnegie Mellon, his research has AI/IR gap. focused on functional Magnetic Resonance

9:00 – 10:30 AM Imaging of the human brain. Mitchell is the Founding Director of CMU’s Center for Auto- IAAI Invited Talk Robot-Assisted Urban Search and Rescue mated Learning and Discovery, an interdisci- at the WTC: Where’s the AI? plinary research center specializing in machine learning and data mining. Robin R. Murphy, Computer Science and Engi- neering, University of South Florida 11:55 AM – 12:45 PM Salon 8, Meeting Level, Shaw Conference Centre AAAI Invited Panel On September 11, 2001, the Center for Robot- AAAI Mobile Robot Competition Assisted Search and Rescue responded within Moderators: Holly Yanco, University of Mas- six hours to the WTC disaster; this is the ﬁrst sachusetts Lowell and Tucker Balch, Georgia Insti- known use of robots for USAR. The Universi- tute of Technology ty of South Florida was one of the four robot Exhibit Hall C, Assembly Level, Shaw Conference Centre teams, and only academic institution. The USF Panelists: Peter Bonasso, Metrica; David Miller, University of Oklahoma and KIPR; Alan Schultz, Naval Research team participated on-site in the search efforts Laboratory; and Reid Simmons, Carnegie Mellon Univer- from September 12 through 22, collecting and sity

archiving data on the use of robots. 11:10 AM – 12:40 PM This talk will provide an overview of the IAAI Invited Panel use of robots for USAR as well as discuss what Pioneering AI Businesses I: AI techniques were available, what was actual- A 20-Year Review ly used, and why. It will also summarize the Moderator: Neil Jacobstein, Chairman, President & key lessons learned from the robotics efforts at CEO of Teknowledge Corporation the WTC. The lessons learned cover the areas Salon 8, Meeting Level, Shaw Conference Centre of platforms and mobility, sensors and sensing Panelists: Ed Feigenbaum, Professor of Computer Science strategies, control, and human-robot interac- and Coscientific Director of the Knowledge Systems Lab- tions. Possibly the most pervasive lesson oratory, Stanford University; Mark Fox, Professor of In- dustrial Engineering, University of Toronto and Chair- learned is that robots for USAR must be con- man & CEO, Novator Systems; Amos Barzilay, General sidered from an “information technology” Partner (United States), Walden International perspective, where platforms, sensors, control Several AI-based businesses started in the schemes, networks, and interfaces must all be early 1980s. They underwent a classic boom 10 and bust cycle. Hype exceeded expectations, learning approach, where action strategies are TalksInvited and some investors and technologists lost pa- inferred from experience. tience. However, history shows that in cases of Different assumptions about the nature of disruptive technological innovation, forecasts actions, sensors, and costs lead to various are usually too optimistic in the short run, and forms of planning: (1) Planning with complete too conservative in the long run. Is that the information and deterministic actions, (2) case with AI businesses? This panel of AI en- planning with non-deterministic actions and trepreneurs will review the technology base sensing, and (3) planning with temporal and and history of pioneering AI businesses, ex- concurrent actions, etc. Most work so far has tract lessons learned, and identify future op- been devoted to “classical” planning (see point portunities. Companies discussed will include 1 above), where significant changes have taken IntelliCorp, Teknowledge, Inference, Syntelli- place in the last few years. On the method- gence, Carnegie Group, Cycorp, and others. An ological side, the area has become more em- interactive question and answer session with pirical with experimental evaluation being panel members will follow brief presentations routine; on the technical side, approaches from each panelist. based on heuristic or constrained-based search have taken over blind-search approaches. 2:00 – 3:30 PM In this talk, Geffner will provide a coherent IAAI-02 Invited Panel picture of planning in AI while trying to con- Pioneering AI Businesses II: Recent Startups vey some of the current excitement in the field. Moderator: Craig Knoblock, Research Associate He’ll make emphasis on the mathematical Professor, University of Southern California, Co- founder and Chief Scientist, Fetch Technologies models that underlie various forms of planning, and the ideas that have been found most Salon 8, Meeting Level, Shaw Conference Centre useful computationally. Panelists: Tom Mitchell, Fredkin Professor of Computer Science, Carnegie Mellon University, Former Chief Scien- tist and VP at WhizBang; Yoav Shoham, Associate Profes- 3:30 – 4:30 PM sor, Stanford University, Co-founder and Chairman, Collocated with the AI Festival TradingDynamics (ARBA) and Cariocas; Daniel Weld, Professor, University of Washington, Co-founder of Net- IAAI-02 AI Entrepreneurs Forum bot, AdRelevance, Nimble Technology, and Asta Net- Exhibit Hall AB, Assembly Level, Shaw Conference works, Venture Partner, Madrona Venture Group Centre This panel will focus on the process of starting This forum will provide an open and informal an AI company. The challenges in creating a setting for AI pioneers, technologists, en- new company include how to apply a technol- trepreneurs, venture capitalists, legal, and in- ogy to address a specific market need and how tellectual property experts to network and dis- to run a successful business. The speakers on cuss issues in starting and running AI-based this panel are AI researchers that have been in- companies. volved in recent startups. Some of the issues the panelists will discuss include how to go Thursday, August 1 from a technology to a business, how to get funding for a company, and what pitfalls to 8:30 – 9:30 AM watch out for. An interactive question and an- AAAI-02/IAAI-02 Joint Invited Talk swer session with panel members will follow Human Level “Strong” AI: The Prospects brief presentations from each panelist. and Implications 2:15 – 3:15 PM Raymond Kurzweil, KurzweilAI.net (Kurzweil Ac- celerating Intelligence Network) AAAI Invited Talk Perspectives on Artificial Intelligence Exhibit Hall C, Assembly Level, Shaw Conference Centre Planning Three-dimensional molecular computing will Hector Geffner, ICREA — Universitat Pompeu provide the hardware for human-level Fabra (Barcelona) “strong”AI well before 2030. The more impor- Exhibit Hall C, Assembly Level, Shaw Conference Centre tant software insights will be gained in part Planning has always been a key area in Artifi- from the reverse-engineering of the human cial Intelligence. In its general form, planning brain, a process well under way. Once nonbio- is concerned with the automatic synthesis of logical intelligence matches the range and sub- action strategies (plans) from a description of tlety of human intelligence, it will necessarily actions, sensors, and goals. Planning thus con- soar past it because of the continuing acceler- trasts with two other approaches to intelligent ation of information-based technologies, as behavior: the programming approach, where well as the ability of machines to instantly action strategies are defined by hand, and the share their knowledge. The implication will be 11 an intimate merger between the technology- analysis of neural coding in several neurobio- creating species and the evolutionary process logical systems that were carried in Tishby’s it spawned. lab. In this talk Tishby will focus on a new approach to data dimensionality reduction that 2:15 – 3:15 PM stems from this principle. Here he searches for AAAI Invited Talk low dimensional (nonlinear) reduction of co- Dimension Reduction that Preserves In- occurrence (or contingency) tables that pre- formation and Neural Coding serve the (mutual) information in the table. Naftali Tishby, The Hebrew University He gives a new alternate-projection algorithm Exhibit Hall C, Assembly Level, Shaw Conference Centre for achieving such a reduction and shows its Many cognitive functions, such as prediction, convergence to an optimal set of information feature extraction, noise filtering, and learn- preserving features. This approach is particu- ing, can be viewed as special cases of one prin- larly useful when the data is not naturally ciple: compression while preserving informa- quantized but rather represented by low dition. This information theoretic principle was mension continuous features. Such a reduc- turned into a computational paradigm: the in- tion may have interesting biological implica- formation bottleneck method. This variation- tions. (Based on joint work with Amir Glober- al method yielded several novel learning and son and Noam Slonim.) data analysis algorithms, with many applica-

Invited Talks tions to information retrieval as well as to Get All the AAAI and IAAI Conference Proceedings, Plus AI Magazine, on CD-ROM SPECIAL RATES!

AAAI 2000 / IAAI 2000 Conference Proceedings CD-ROM This CD-ROM set consists of all the papers from the 2000 National Conferences on Artiﬁcial Intelligence, as well as the 2000 Innovative Ap- plications of Artiﬁcial Intelligence Conferences. The single CD contains approximately 260 technical papers in PDF format, and is readable by most Windows (1998 and later), Macintosh, and many UNIX platforms. 1 CD ($45.00 CDN — AAAI members; $175.00 CDN nonmembers)

AAAI 1998–99, IAAI 1998–99, KDD 1995–98 CD-ROM Set This CD-ROM set consists of all the papers from the 1998 and 1999 National Conferences on Artiﬁcial Intelligence, as well as the 1998 and 1999 Innovative Applications of Artiﬁcial Intelligence Conferences. Included also on this CD are the 1995, 1996, 1997, and 1998 Knowledge Discovery and Data Mining Conference papers. The single CD contains approximately 260 technical papers in PDF format, and is readable by most Windows, Macintosh, and many UNIX platforms. 1 CD ($45.00 CDN — AAAI members; $175.00 CDN nonmembers)

AAAI 1980–97, IAAI 1993–97 CD-ROM Set This archival CD-ROM set consists of all the papers from the 1980, 1982, 1983, 1984, 1986, 1987, 1988, 1990, 1991, 1992, 1993, 1994, 1996, and 1997 AAAI National Conferences, as well as the papers from the 1993, 1994, 1995, 1996, and 1997 Innovative Applications of Artiﬁ- cial Intelligence Conferences. The set consists of more than 2,200 papers in PDF format, and is readable by most Windows, Macintosh, and many UNIX platforms. 1 CD ($95.00 CDN — AAAI members; $250.00 CDN nonmembers)

AI Magazine 1980–2001 CD-ROM Set This new CD-ROM consists of all the articles from AI Magazine published between 1980 and 2001. The single CD is readable by most Win- dows (1998 and later), Macintosh, and many UNIX platforms. 1 CD ($45.00 CDN — AAAI members; $175. CDN nonmembers) To purchase, see the AAAI Conference Registrar. You must be an AAAI member in good standing, or purchase an AAAI membership onsite to qualify for the conference special membership rate. Offer expires August 1, 2002. Purchase must be made at AAAI–02 onsite. No additional discounts apply.

12 Tuesday, July 30—AAAI-02 Sessions Tuesday, July 30—IAAI-02 Sessions (All AAAI-02 sessions will be held in Hall C.) (All IAAI-02 sessions will be held in Salon 8, unless otherwise noted.) : - :  : - :  AAAI-02 Opening Remarks and Outstanding Paper Award Presentation AAAI-02 Opening Remarks and Outstanding Paper Award Presentation Rina Dechter and Michael Kearns, AAAI-02 Program Cochairs Rina Dechter and Michael Kearns, AAAI-02 Program Cochairs IAAI-02 Awards Presentation IAAI-02 Awards Presentation Steve Chien and John Riedl, IAAI-02 Chair and Cochair Steve Chien and John Riedl, IAAI-02 Chair and Cochair (Hall C) : - :  Keynote Address : - :  Probabilistic AI and Information Retrieval IAAI-02 Opening Remarks Michael I. Jordan, University of California, Berkeley Steve Chien and John Riedl, IAAI-02 Chair and Cochair Introduction: Rina Dechter and Michael Kearns : - :  : - :  Invited Talk Probabilistic Reasoning Robot Assisted Urban Search and Rescue at the WTC: Session Chair: David McAllester Where’s the AI? A Distance Measure for Bounding Probabilistic Belief Change Robin Murphy, University of South Florida Hei Chan and Adnan Darwiche Introduction: David Kortenkamp Visual Exploration and Incremental Utility Elicitation : - :  Jim Blythe Break A Graphical Criterion for the Identiﬁcation of Causal Effects in Linear Models Carlos Brito and Judea Pearl :  - :  Perception and Monitoring : - :  Session Chair: David Kortenkamp Break Applying Perceptually Driven Cognitive Mapping to :  - :  Virtual Urban Environments (Emerging Technology) Randall W. Hill, Jr., Changhee Han, and Michael van Lent Planning MiTAP, Text and Audio Processing for Bio-Security: A Case Study Session Chair: Weixiong Zhang (Deployed Application) Plan Evaluation with Incomplete Action Descriptions Laurie Damianos, Jay Ponte, Steve Wohlever, Florence Reeder, David Day, Andrew Garland and Neal Lesh George Wilson, and Lynette Hirschman Symbolic Heuristic Search for Factored Markov Decision Processes Computational Vulnerability Analysis for Information Survivability Zhengzhu Feng and Eric A. Hansen (Emerging Technology) Algorithms for a Temporal Decoupling Problem in Multi-Agent Planning Howard Shrobe Luke Hunsberger : - :  : - :  Lunch Lunch Break : - :  AAAI-02/IAAI-02 Program Committee Lunch (by invitation only) Planning and Scheduling Session Chair: Steve Chien : - :  Staff Scheduling for Inbound Call Centers and Customer Contact Centers Learning I (Deployed Application) Session Chair: Peter Stone Alex Fukunaga, Ed Hamilton, Jason Fama, David Andre, Ofer Matan, and Illah Nourbakhsh Polynomial-Time Reinforcement Learning of Near-Optimal Policies Karèn Pivazyan and Yoav Shoham Getting from Here to There: Interactive Planning and Agent Execution for Optimizing Travel (Emerging Technology) Minimum Majority Classiﬁcation and Boosting José Luis Ambite, Greg Barish, Craig A. Knoblock, Philip M. Long Maria Muslea, Jean Oh, and Steven Minton Data Perturbation for Escaping Local Maxima in Learning UTTSExam: A Campus-Wide University Exam-Timetabling System Gal Elidan, Matan Ninio, Nir Friedman, and Dale Shuurmans (Deployed Application) Andrew Lim, Juay-Chin Ang, Wee-Kit Ho, and Wee-Chong Oon : - :  Break : - :  : - :  Break Constraint Satisfaction/Search : - :  Session Chair: Weixiong Zhang Knowledge Formation and Management A New Algorithm for Optimal Bin Packing Session Chair: Bob Engelmore Richard E. Korf Knowledge Formation and Dialogue Using the KRAKEN Toolset PROMPTDIFF: A Fixed-Point Algorithm for Comparing Ontology Versions (Emerging Technology) Natalya F. Noy and Mark A. Musen Kathy Panton, Pierluigi Miraglia, Nancy Salay, Robert C. Kahlert, Node and Arc Consistency in Weighted CSP David Baxter, and Roland Reagan Javier Larrosa A Web-Based Ontology Browsing and Editing System (Emerging Technology) Model Induction: A New Source of CSP Model Redundancy Jérôme Thoméré, Ken Barker, Vinay Chaudhri, Peter Clark, Michael Eriksen, Y.C. Law and J.H.M. Lee Sunil Mishra, Bruce Porter, and Andres Rodriguez WhyNot: Debugging Failed Queries in Large Knowledge Bases : ‒ :  (Emerging Technology) Poster Session I Hans Chalupsky and Thomas A. Russ Ballroom, Westin Edmonton

13 Wednesday, July 31—AAAI-02 Sessions Wednesday, July 31—IAAI-02 Sessions :  :  AAAI Special Awards Presentation AAAI Special Awards Presentation Tom Mitchell, AAAI President and Bruce Buchanan, Tom Mitchell, AAAI President and Bruce Buchanan, AAAI Past President and Awards Committee Chair AAAI Past President and Awards Committee Chair (Hall C) : - :  AAAI Presidential Address : - :  AI and the Impending Revolution in Brain Science AAAI Presidential Address Tom M. Mitchell, Carnegie Mellon University AI and the Impending Revolution in Brain Science Introduction: Bruce G. Buchanan, AAAI Past President Tom M. Mitchell, Carnegie Mellon University Introduction: Bruce G. Buchanan, AAAI Past President : - :  (Hall C) Representation and Reasoning / Robot Application Session Chair: Richard Korf : - :  AAAI-02 Outstanding Paper Learning and Design On Computing all Abductive Explanations Session Chair: John Riedl Thomas Eiter and Kazuhisa Makino A Structure Based Conﬁguration Tool: Drive Solution Designer – DSD Measuring Inconsistency in Knowledge via Quasi-Classical Models (Deployed Application) Anthony Hunter Christoph Ranze Thorsten Scholz, Thomas Wagner, Andreas Günter, FastSLAM: A Factored Solution to the Simultaneous Localization Otthein Herzog, Oliver Hollmann, Christoph Schlieder, and Volker Arlt and Mapping Problem An Analogy Ontology for Integrating Analogical Processing and Michael Montemerlo, Sebastian Thrun, Daphne Koller, and Ben Wegbreit First-Principles Reasoning (Emerging Technology) Kenneth D. Forbus, Thomas Mostek, and Ron Ferguson : - :  Break : - :  :  - :  Break Robot Applications :  - :  Session Chair: Dieter Fox Invited Panel Watch Their Moves: Applying Probabilistic Multiple Object Tracking Pioneering AI Businesses I: A 20-Year Review to Autonomous Robot Soccer Moderator: Neil Jacobstein, Teknowledge Corporation Thorsten Schmitt, Michael Beetz, Robert Hanek, and Sebastian Buck 12:40 – 2:00 pm Experiences with a Mobile Robotic Guide for the Elderly Lunch Michael Montemerlo, Joelle Pineau, Nicholas Roy, Sebastian Thrun, and Vandi Verma : ‒ :  Invited Panel :  - :  Pioneering AI Businesses II: Recent Startups Invited Panel Moderator: Craig Knoblock, University of Southern California and AAAI Mobile Robot Competition Fetch Technologies Moderators: Holly Yanco, University of Massachusetts Lowell and Tucker Balch, Georgia Institute of Technology : - :  Introduction: Michael Kearns AI Festival Panelists: Peter Bonasso, Metrica; David Miller, University of Oklahoma (Hall AB) and KIPR; Alan Schultz, Naval Research Laboratory; and : - :  Reid Simmons, Carnegie Mellon University IAAI-02 AI Entrepreneurs Forum Collocated with AI Festival : - :  Lunch Break : - :  Invited Talk Perspectives on Artiﬁcial Intelligence Planning Hector Geffner, ICREA - Universitat Pompeu Fabra (Barcelona) Introduction: Rina Dechter : - :  AI Festival : - :  Poster Session II Ballroom, Westin Edmonton

14 Thursday, August 1—AAAI-02 Sessions Thursday, August 1—IAAI-02 Sessions : - :  : - :  AAAI-02/IAAI-02 Joint Invited Talk AAAI-02/IAAI-02 Joint Invited Talk Human Level "Strong" AI: The Prospects and Implications Human Level "Strong" AI: The Prospects and Implications Raymond Kurzweil, KurzweilAI.net (Kurzweil Accelerating Intelligence Network) Raymond Kurzweil, KurzweilAI.net (Kurzweil Accelerating Intelligence Network) Introduction: Neil Jacobstein Introduction: Neil Jacobstein (Hall C) : - :  Learning II : - :  Session Chair: David McAllester Kurzweil Q&A Session (Salon 8) State Abstraction for Programmable Reinforcement Learning Agents David Andre and Stuart J. Russell : - :  Learning Temporal, Relational, Force-Dynamic Event Definitions from Video Break Alan Fern, Jeffrey Mark Siskind, and Robert Givan Bootstrap Learning for Place Recognition :  - :  Benjamin Kuipers and Patrick Beeson Internet Session Chair: Neil Jacobstein : - :  The 2001 Trading Agent Competition (Emerging Technology) Break Michael P. Wellman, Amy Greenwald, Peter Stone, and Peter R. Wurman :  - :  RightNow eService Center: Internet Customer Service Using a Self-Learning Knowledge Base (Deployed Application) Multiagent Systems Stephen D. Durbin, Doug Warner, J. Neal Richter, and Zuzana Gedeon Session Chair: Peter Stone A Decision-Support System for Quote Generation (Deployed Application) (Im)possibility of Safe Exchange Mechanism Design Richard Goodwin, Rama Akkiraju, and Fred Wu Tuomas Sandholm and XiaoFeng Wang Multi-Agent Algorithms for Solving Graphical Games : - :  David Vickrey and Daphne Koller Lunch Reinforcement Learning of Coordination in Cooperative Multi-Agent Systems Spiros Kapetanakis and Daniel Kudenko : - :  Complexity of Manipulating Elections with Few Candidates Military Applications Vincent Conitzer and Tuomas Sandholm Session Chair: Russell Knight Toward Practical Knowledge-Based Tools for Battle Planning and Scheduling : - :  (Emerging Technology) Lunch Alexander Kott, Larry Ground, Ray Budd, Lakshmi Rebbapragada, and John Langston : - :  AI on the Battlefield: An Experimental Exploration (Emerging Technology) Invited Talk Robert Rasch, Alexander Kott, and Kenneth D. Forbus Dimension Reduction that Preserves Information and Neural Coding Development and Deployment of a Disciple Agent for Center of Gravity Naftali Tishby, The Hebrew University Analysis (Deployed Application) Introduction: Michael Kearns Gheorghe Tecuci, Mihai Boicu, Dorin Marcu, Bogdan Stanescu, Christina Boicu, Jerry Comello, Antonio Lopez, James Donlon, and William Cleckner : - : Intelligent Control of Auxiliary Ship Systems (Emerging Technology) Break David Scheidt, Christopher McCubbin, Michael Pekala, Shon Vick, : - :  and David Alger Satisfiability Session Chair: Tuomas Sandholm Enhancing Davis Putnam with Extended Binary Clause Reasoning Fahiem Bacchus Comparing Phase Transitions and Peak Cost in PP-Complete Satisfiability Problems Delbert D. Bailey, Víctor Dalmau, and Phokion G. Kolaitis Learning for Quantified Boolean Logic Satisfiability Enrico Giunchiglia, Massimo Narizzano, and Armando Tacchella Dynamic Restart Policies Henry Kautz, Eric Horvitz, Yongshao Ruan, Carla Gomes, and Bart Selman

15 CobotDS: A Spoken Dialogue System for Chat Poster Session I Michael Kearns, Charles Isbell, Satinder Singh, Diane Litman, and Tuesday, July 29 (5:45 - 8:45 PM) Jessica Howe Exploiting Auditory Fovea in Humanoid-Human Interaction Constraint Satisfaction / Search Kazuhiro Nakadai, Hiroshi G. Okuno, and Hiroaki Kitano The Yard Allocation Problem Towards CST-Enhanced Summarization Ping Chen, Zhaohui Fu, and Andrew Lim Zhu Zhang, Sasha Blair-Goldensohn, and Dragomir R. Radev Searching for Backbones and Fat: A Limit-Crossing Approach with Applications Planning Sharlee Climer and Weixiong Zhang Planning with a Language for Extended Goals Generating Random Solutions for Constraint Satisfaction Problems Ugo Dal Lago, Marco Pistore, and Paolo Traverso Rina Dechter, Kalev Kask, Eyal Bin and Roy Emek Symbolic Heuristic Search for Factored Markov Graph Coloring with Quantum Heuristics Decision Processes Alex Fabrikant and Tad Hogg Zhengzhu Feng and Eric A. Hansen Preference-Based Search and Multi-Criteria Optimization Plan Evaluation with Incomplete Action Descriptions Ulrich Junker Andrew Garland and Neal Lesh Human-Guided Tabu Search Algorithms for a Temporal Decoupling Problem in Gunnar W. Klau, Neal Lesh, Joe Marks, and Michael Mitzenmacher Multi-Agent Planning Node and Arc Consistency in Weighted CSP Luke Hunsberger Javier Larrosa D*Lite Integrating Local Search and Network Flow to Solve the Sven Koenig and Maxim Likhachev Inventory Routing Problem Speeding Up the Calculation of Heuristics for Hoong Chuin Lau, Qizhang Liu, and Hirotaka Ono Heuristic Search-Based Planning Model Induction: A New Source of CSP Model Redundancy Yaxin Liu, Sven Koenig and David Furcy Y.C. Law and J.H.M. Lee Iterative-Refinement for Action Timing Discretization Todd W. Neller Learning A Logical Measure of Progress for Planning Contentful Mental States for Robot Baby Aarati Parmar Paul R. Cohen, Tim Oates, Carole R. Beal, and Niall Adams Data Perturbation for Escaping Local Maxima in Learning Probabilistic and Causal Reasoning Gal Elidan, Matan Ninio, Nir Friedman, and Dale Shuurmans Visual Exploration and Incremental Utility Elicitation Pruning and Dynamic Scheduling of Cost-Sensitive Ensembles Jim Blythe Wei Fan, Fang Chu, Haixun Wang, and Philip S. Yu A Graphical Criterion for the Identification of Causal Effects in Specific-to-General Learning for Temporal Events Linear Models Alan Fern, Robert Givan, and Jeffrey Mark Siskind Carlos Brito and Judea Pearl Structural Extension to Logistic Regression: Discriminative A Distance Measure for Bounding Probabilistic Belief Change Parameter Learning of Belief Net Classifiers Hei Chan and Adnan Darwiche Russell Greiner and Wei Zhou Strategies for Determining Causes of Events Bootstrap Learning for Place Recognition Mark Hopkins Benjamin Kuipers and Patrick Beeson Tree Approximation for Belief Updating Reinforcement Learning for POMDPs Based on Action Values Robert Mateescu, Rina Dechter, and Kalev Kask and Stochastic Optimization Accuracy vs. Efficiency Trade-offs in Probabilistic Diagnosis Theodore J. Perkins Irina Rish, Mark Brodie, and Sheng Ma Polynomial-Time Reinforcement Learning of Near-Optimal Policies A General Identification Condition for Causal Effects Karèn Pivazyan and Yoav Shoham Jin Tian and Judea Pearl Hierarchical Latent Class Models for Cluster Analysis A New Characterization of the Experimental Implications of Nevin L. Zhang Causal Bayesian Networks Jin Tian and Judea Pearl Markov Decision Processes A POMDP Formulation of Preference Elicitation Problems Satisfiability Craig Boutilier Enhancing Davis Putnam with Extended Binary Clause Reasoning Context-Specific Multiagent Coordination and Planning with Fahiem Bacchus Factored MDPs Comparing Phase Transitions and Peak Cost in PP-Complete Carlos Guestrin, Shobha Venkataraman, and Daphne Koller Satisfiability Problems Nearly Deterministic Abstractions of Markov Decision Processes Delbert D. Bailey, Víctor Dalmau, and Phokion G. Kolaitis Terran Lane and Leslie Pack Kaelbling Inference Methods for a Pseudo-Boolean Satisfiability Solver

Technical Poster SessionsThe Size of MDP Factored Policies Heidi E. Dixon and Matthew L. Ginsberg Paolo Liberatore Automated Discovery of Composite SAT Variable-Selection Greedy Linear Value-Approximation for Factored Heuristics Markov Decision Processes Alex Fukunaga Relu Patrascu, Pascal Poupart, Dale Schuurmans, Craig Boutilier Learning for Quantified Boolean Logic Satisfiability and Carlos Guestrin Enrico Giunchiglia, Massimo Narizzano, and Armando Tacchella An Adaptive Noise Mechanism for WalkSAT Multiagent Systems I Holger H. Hoos Complexity of Manipulating Elections with Few Candidates Vincent Conitzer and Tuomas Sandholm Vision A Logic-Based Model of Intentions for Multi-Agent Subcontracting Detection and Classification of Motion Boundaries John Grant, Sarit Kraus, and Donald Perlis Richard Mann and Allan D. Jepson Reinforcement Learning of Coordination in Recognizing Multitasked Activities from Video Using Stochastic Cooperative Multi-Agent Systems Context-Free Grammar Spiros Kapetanakis and Daniel Kudenko Darnell Moore and Irfan Essa The Design of Collectives of Agents to Control The OD Theory of TOD: The Use and Limits of Temporal Non-Markovian Systems Information for Object Discovery John W. Lawson and David H. Wolpert Brandon C. S. Sanders, Randal C. Nelson, and Rahul Sukthankar (Im)possibility of Safe Exchange Mechanism Design Tuomas Sandholm and XiaoFeng Wang Multi-Agent Algorithms for Solving Graphical Games David Vickrey and Daphne Koller Poster Session II Distributed Breakout Revisited Wednesday, July 30 (7:00 - 10:00 PM) Weixiong Zhang and Lars Wittenburg Constraint Satisfaction / Search Scheduling Contract Algorithms on Multiple Processors Natural Language Processing Daniel S. Bernstein, Theodore J. Perkins, Shlomo Zilberstein, and Learning Pattern Rules for Chinese Named Entity Extraction Lev Finkelstein Tat-Seng Chua and Jimin Liu Multiple-Goal Search Algorithms and their Application to Web Language Modeling for Soft Keyboards Crawling Joshua Goodman, Gina Venolia, Keith Steury, and Chauncey Parker Dmitry Davidov and Shaul Markovitch 16 Reducing Search Space in Local Search for Constraint Satisfaction Sven Wachsmuth and Gerhard Sagerer Presentations Poster Technical H. Fang, Y. Kilani, J.H.M. Lee, and P.J. Stuckey Value Iteration Working with Belief Subset Optimal Schedules for Parallelizing Anytime Algorithms: The Weixiong Zhang and Nevin L. Zhang Case of Independent Processes Lev Finkelstein, Shaul Markovich, and Ehud Rivlin Multiagent Systems II: Auctions Optimal Depth-First Strategies for And-Or Trees Russell Greiner, Ryan Hayward, and Michael Molloy Solving Concisely Expressed Combinatorial Auction Problems Craig Boutilier A New Algorithm for Optimal Bin Packing Richard E. Korf Partial-Revelation VCG Mechanism for Combinatorial Auctions Wolfram Conen and Tuomas Sandholm Memory-Efficient A* Heuristics for Multiple Sequence Alignment Matthew McNaughton, Paul Lu, Jonathan Schaeffer, and Duane Bidding Clubs in First-Price Auctions Szafron Kevin Leyton-Brown, Yoav Shoham, and Moshe Tennenholtz PROMPTDIFF: A Fixed-Point Algorithm for Comparing Truthful Approximation Mechanisms for Restricted Combinato- Ontology Versions rial Auctions Natalya F. Noy and Mark A. Musen Ahuva Mu’alem and Noam Nisan On Preference-Based Search in State Space Graphs Structural Leverage and Fictitious Play in Sequential Auctions Patrice Perny and Olivier Spanjaard Weili Zhu and Peter R. Wurman An Average-Case Analysis of Graph Search Anup K. Sen, Amitava Bagchi, and Weixiong Zhang Multiagent Systems III: Game Theory Vote Elicitation: Complexity and Strategy-Proofness Knowledge Representation I Vincent Conitzer and Tuomas Sandholm On Computing all Abductive Explanations Dispersion Games: General Definitions and Some Thomas Eiter and Kazuhisa Makino Specific Learning Results / Trond Grenager, Rob Powers, and Yoav Shoham Measuring Inconsistency in Knowledge via Quasi-Classical Models Anthony Hunter Competitive Safety Analysis Moshe Tennenholtz A Hoare-Style Proof System for Robot Programs Yongmei Liu Representing and Reasoning about Mappings between Planning II: Actions & Temporal Reasoning Domain Models Reasoning about Actions in a Probabilistic Setting Jayant Madhavan, Philip A. Bernstein, Pedro Domingos, Chitta Baral, Nam Tran, and Le-Chi Tuan and Alon Y. Halevy A Method for Metric Temporal Reasoning A Regression Based Adaptation Strategy for Case-Based Reasoning Mathias Broxvall David Patterson, Niall Rooney, and Mykola Galushka Non-Markovian Control in the Situation Calculus Cluster Ensembles – A Knowledge Reuse Framework for Alfredo Gabaldon Combining Partitionings Alexander Strehl and Joydeep Ghosh Robotics Robust Global Localization Using Clustered Particle Filtering Knowledge Representation II: Adam Milstein, Javier Nicolás Sánchez, and Evan Tang Williamson Experiences with a Mobile Robotic Guide for the Elderly Logic Programming Michael Montemerlo, Joelle Pineau, Nicholas Roy, Sebastian Thrun, Logic Programming with Ordered Disjunction and Vandi Verma Gerhard Brewka FastSLAM: A Factored Solution to the Simultaneous Localization A Three-Valued Characterization for Strong Equivalence of Logic and Mapping Problem Programs Michael Montemerlo, Sebastian Thrun, Daphne Koller, and Pedro Cabalar Ben Wegbreit ASSAT: Computing Answer Sets of a Logic Program by SAT Watch Their Moves: Applying Probabilistic Multiple Object Solvers Tracking to Autonomous Robot Soccer Fangzhen Lin and Yuting Zhao Thorsten Schmitt, Michael Beetz, Robert Hanek, and Sebastian Buck Learning CD*: A Real-Time Resolution Optimal Re-Planner for Globally State Abstraction for Programmable Reinforcement Learning Constrained Problems Agents Anthony Stentz David Andre and Stuart J. Russell Progressive Rademacher Sampling Satisfiability Tapio Elomaa and Matti Kääriäinen A Compiler for Deterministic, Decomposable Negation Normal Learning Temporal, Relational, Force-Dynamic Event Definitions Form from Video Adnan Darwiche Alan Fern, Jeffrey Mark Siskind, and Robert Givan A Mixture-Model for the Behaviour of SLS Algorithms for SAT Minimum Majority Classification and Boosting Holger H. Hoos Philip M. Long SetA*: An Efficient BDD-Based Heuristic Search Algorithm Content-Boosted Collaborative Filtering for Rune M. Jensen, Randal E. Bryant, and Manuela M. Veloso Improved Recommendations Dynamic Restart Policies Prem Melville, Raymond J. Mooney, and Ramadass Nagarajan Henry Kautz, Eric Horvitz, Yongshao Ruan, Carla Gomes, Constructive Adaptive User Interfaces – Composing Music and Bart Selman Based on Human Feelings Using Weighted MAX-SAT Engines to Solve MPE Masayuki Numao, Shoichi Takagi, and Keisuke Nakamura James D. Park Constrained Formulations and Algorithms for Stock-Price Easy Predictions for the Easy-Hard-Easy Transition Predictions Using Recurrent FIR Neural Networks Andrew J. Parkes Benjamin W. Wah and Minglun Qian The Interface between P and NP: COL, XOR, NAE, 1-in-k, and Rule-Based Anomaly Pattern Detection for Detecting Disease Horn SAT Outbreaks Toby Walsh Weng-Keen Wong, Andrew Moore, Gregory Cooper, and Michael Wagner Extended Isomap for Pattern Classification Web and Information Extraction Ming-Hsuan Yang A Maximum Entropy Approach to Information Extraction from Semi-Structured and Free Text Hai Leong Chieu and Hwee Tou Ng Markov Decision Processes Reviewing the Design of DAML+OIL: An Ontology Language for Segmenting Time Series with a Hybrid Neural Networks — the Semantic Web Hidden Markov Model Ian Horrocks, Peter F. Patel-Schneider, and Frank van Harmelen Laura Firoiu and Paul R. Cohen Stochastic Link and Group Detection On Policy Iteration as a Newton’s Method and Polynomial Policy Jeremy Kubica, Andrew Moore, Jeff Schneider, and Yiming Yang Iteration Algorithms Omid Madani Efficient Utility Functions for Ceteris Paribus Preferences Michael McGeachie and Jon Doyle Piecewise Linear Value Function Approximation for Factored MDPs Pascal Poupart, Craig Boutilier, Relu Patrascu, and Dale Schuurmans Bayesian Networks for Speech and Image Integration 17 Exhibition Email: [email protected] Web: www.activmedia.com The exhibition will be held in Exhibit Hall AB ActivMedia Robots team of compatible sys- on the assembly level of the Shaw Conference tems offers various scales, terrain capabilities, Centre, Tuesday, July 30 and Wednesday, July payloads and human interfaces, used by thou- 31. Admittance is restricted to badged confer- sands of roboticists around the globe. These ence attendees. Vendor-issued guest passes cross-Linux/WIN platforms feature powerful must be redeemed at the Exhibitor Registra- ARIA developer software plus integrated map- tion Desk, in the registration area, on the as- ping, Saphira gradient navigation, scheduling, sembly level of the Shaw Conference Centre. and multitude of plug-n-play accessories. We Further information regarding access to the take care of the basics so you can focus on Exhibition can be obtained from the Exhibitor what’s new and exciting in your research. Ac- Registration Desk. Guest tickets to the AI Fes- tivMedia Robots accessories include inertial tival can be purchased for $25.00 CDN per correction, integrated laser mapping and nav- person ($8.00 CDN for children) at onsite reg- igation, active PTZ vision, color-tracking and istration. surveillance, GPS, tilt-skew, bumpers, grip- pers, fixed IR, arms, active PTZ stereovision, Exhibit Hours sonar and more. Come see us in action!

Tuesday, July 30: 10:00 AM – 5:30 PM Booth # 203 Wednesday, July 31: 10:00 AM – 3:00 PM AI TOPICS — the AAAI Pathfinder 3:00 PM – 5:30 PM (AI Festival) Webmaster: Jon Glick Website: www.aaai.org/aitopics Exhibitors E-mail: [email protected] AI TOPICS has something for everyone. Hav- ■ Exhibit Program Exhibit AAAI Press ing started as a dynamic online library spon- ■ ActivMedia Robotics sored by the American Association for Artifi- ■ AI Topics cial Intelligence for students, teachers, jour- ■ Applied AI Systems, Inc. nalists, and others in need of basic, introduc- ■ Elsevier Science tory information about what artificial intelli- ■ Franz Inc. gence is and what AI scientists do, our recent ■ iCORE expansion of the “AI in the news” feature has ■ iRobot Corp., Research Robots Division broadened the site’s scope such that everyone ■ Kluwer Academic Publishers who is interested in the exciting field of artifi- ■ KurzweilAI.net cial intelligence now has a great reason to vis- ■ The MIT Press it AI TOPICS. And because we are dedicated ■ Morgan Kaufmann Publishers to serving you, we hope that you take advan- ■ NASA Ames Research Center tage of this opportunity to stop by Booth #203 ■ Sony Electronics Inc., Entertainment to take a guided tour of the website, to share Robotic America your comments, concerns, ideas and sugges- ■ Springer-Verlag New York, Inc. tions, or to just catch up on the latest AI news.

Booth #100 Booth #309 AAAI Press Applied AI Systems, Inc. 445 Burgess Drive 3232 Carp Road, RR 2 Menlo Park, CA 94025-3442 Ontario, Canada K0A 1L0 Tel: 650-328-3123 Fax: 650-321-4457 Applied AI Systems, Inc. (AAI) will be demon- Email: [email protected] strating Khepera II, Koala, and GAIA robots Online Catalog: www.aaaipress.org/ and Webots simulation software, along with AAAI Press is the only press headed by prac- many options and accessories. Discuss new ticing AI scientists. We are pleased to offer a features of recently released Khepera II with 20% discount to conference attendees. knowledgeable engineers who are users of the robots. Videos of other larger robots and their Booth #208 applications such as transportation for the el- ActivMedia Robotics derly, intelligent wheelchairs for the disabled, 44 Concord Street movement of heavy materials autonomously Peterborough, NH 03458 in construction or farming industry, and intel- Tel: 603-924-9100 ligent camera to detect and warn of dangerous 18 situations (e.g., landslide or illegal behavior sition Alberta as a leader in information and Exhibit Program (e.g., garbage dumping) will be shown). communications technology by establishing and funding top research teams. Booth #114 Ten new exceptional chairs and their re- Elsevier Science search teams are well under way. iCORE in- 655 Avenue of the Americas tends to fund the establishment of over a New York, NY 10010 dozen more new chairs at Alberta universities Elsevier Science is the market leader of scien- over the next several years. iCORE also sup- tific publications. All books and journals in ports nearly 90 ICT-related graduate students the field of artificial intelligence, expert sys- at universities around Alberta through a sepa- tems and knowledge-based systems will be on rate scholarship program. display. Also available from our booth, some of Elsevier’s electronic initiatives including the Booth #207 Computer Science Preprint Server (www. iRobot Corporation, computersciencepreprints.com)—the perma- Research Robots Division nent web archive and rapid distribution medi- 37 Wilton Road um for research articles in the field of com- Milford, NH 03055 puter science—and the new to launch Com- 603-654-3400 puter Science Portal (www.computer- www.irobot.com scienceweb.org). ComputerScienceWeb offers Inspire. Encourage. Enable. full text articles from leading computer sci- iRobot’s team of dedicated engineers, software ence journals and integrated services to facili- developers and production specialists em- tate research in all fields of computer science. brace creativity innovative thinking and cut- ting-edge technology to design and build a Booth #221 growing family of versatile, rugged, fully inte- Franz Inc. grated mobile robot systems. The revolution- ary Mobility™ Robot Integration Software 1995 University Avenue Ste #275 and rFLEX™ Robot Control Architecture Berkeley, CA 94704 drive the entire family of research robot plat- Franz Inc. produces Allegro CL® 6.2, a com- forms, providing seamless, top-to-bottom in- plete, cross-platform development environ- tegration along with clear, intuitive migration ment powered by Common Lisp/CLOS. Alle- paths among platforms. iRobot’s mission is to gro CL’s dynamic object-oriented technology provide robot development tools that inspire, allows developers to create leading edge, mis- encourage and enable advances in robotics. sion-critical applications that are robust, Stop by to see these great red machines in ac- scaleable, and easy to evolve and deploy. Alle- tion. gro CL is ideal for Dynamic Servers, Manufac- turing scheduling and control, IC design & Booth #108 synthesis, Knowledge Management and Data Kluwer Academic Publishers Mining. Other Franz Inc. products include Al- 101 Philip Drive legroServe®, a dynamic web-enabling Lisp Norwell, MA 02061 Web Server, Allegro ORBLink®, a CORBA- Kluwer Academic Publishers, a leading pub- compliant ORB, and AllegroStore®, a persis- lisher of scientific books and journals, invites tent object database. attendees to visit our display and browse through our latest publications. Pick up a Booth #106 sample copy of our wide array of artificial in- iCORE telligence journals including Machine Learn- Contact: Mary Anne Moser ing, Autonomous Robots, and Data Mining and 3608 33 Street NW Knowledge Discovery. Also, receive a 20% dis- Calgary, AB, T2L 2A6 Canada count on all Kluwer titles on display. For more Tel: 403-949-3306 details on all our publications visit www. Fax: 403-949-3320 wkap.nl email: [email protected] The Informatics Circle of Research Excellence Booth #120 (iCORE) was created three years ago to develop and support excellent university-based re- KurzweilAI.net search in information and communications 15 Walnut St. technology (ICT) in Alberta. Its goal is to po- Suite 200 Wellesley Hills, MA 02481 19 KurzweilAI.net is the web home of today’s big The NASA booth features information on the thinkers and newsmakers examining the con- development of autonomous technologies for fluence of accelerating revolutions shaping robotic spacecraft and an overview of NASA’s our future world and the inside story on new Intelligent Systems Program. The IS Program technological and social realities. It focuses es- directs research in automated reasoning, hu- pecially on the exponential growth of intelli- man-centered computing and intelligent data gence, both biological and machine, and the understanding that will lead to the advanced merger of the two in a post-humanist future. computer science and information technolo- Visit us on the web at KurzweilAI.net gies essential to NASA’s missions of exploration in the 21st Century. The NASA booth Booth #102 will also feature a demonstration of the Per- The MIT Press sonal Rover Project, a collaborative effort be- Five Cambridge Center tween NASA and Carnegie Mellon University. Cambridge, MA 02142 Tel: 800-56-0343 Booth #215 Fax: 617-253-1709 Sony Electronics Inc., Email: [email protected] Web: mitpress.mit.edu Entertainment Robotic America MIT Press is one of the leading publishers in Contact: John DeCuir the field of artificial intelligence. We are 16450 West Bernardo Drive pleased to offer a wide range of our recent San Diego, CA 92127 Tel: 858-942-3818 publications at a conference discount of 20%. Fax: 858-942-9080 Sony is pleased to offer the OPEN-R™ SDK Booth #112 (Software Development Kit) for AIBO® Enter- Morgan Kaufmann Publishers tainment Robots. OPEN-R is the standard in-

Exhibit Program Exhibit An Imprint of Elsevier Science and Technology terface for Sony’s entertainment robot system. Books This interface greatly expands the capabilities 340 Pine Street, 6th Floor of entertainment robots. The OPEN-R SDK is San Francisco, CA 94104 Tel: 415-392-2665 a cross-development environment based on Fax: 415-982-2665 GCC (C++) which you can use to make non- www.mkp.com commercial software applications for AIBO Morgan Kaufmann is dedicated to publishing (ERS-210, ERS-220, ERS-210A, and ERS- distinguished books for artificial intelligence 220A). The tools are free of charge! Please vis- researchers and students. Our list of AI publi- it us at www.aibo.com/ openr/ to access tools, cations reflects the classical disciplines within documentation, and a bulletin board where AI including representation and reasoning, you can collaborate with other developers. machine learning, natural language processing, intelligent systems, computer vision, Booth #213 planning and uncertainty as well as emerging Springer-Verlag New York, Inc. areas such as agents, data mining and evolu- 175 Fifth Avenue tionary computation. We publish textbooks, New York, New York 10010 USA surveys of areas and methods, programming Tel: 800-777-4643 guides, and select research volumes and pro- [email protected] ceedings. New, just in time for AAAI 2002, Ex- www.springer-ny.com ploring Artificial Intelligence in the New Millen- Save 20% on artificial intelligence books from nium. Springer-Verlag! From the Lecture Notes in Computer Science series to the upcoming title Booth #200 Agent Technology, Springer has established a NASA Ames Research Center reputation for publishing essential books and resources in artificial intelligence and agents. Contact: Robert Duffy M/S 269-1 Moffett Field, CA 94035-1000 Tel: 650-604-3591 The Information Sciences Directorate of NASA Ames Research Center is a world-class creator of innovative, intelligent, high performance and reliable information technologies for NASA’s space and aeronautics missions. 20 Intelligent Systems Demos Intelligent 4:00 PM:Disciple-RKF/COG: Agent Teaching Intelligent Systems by Subject Matter (Booth 413) Gheorghe Demonstrations Tecuci The Intelligent Systems Demonstrations will Wednesday, July 31 be held in Exhibit Hall AB of the Shaw Con- 10:00 AM:Multi-ViewPoint Clustering Analy- ference Centre and will be open to registered sis Tool (Booth D318) Mala Mehrotra Pra- conference attendees during exhibit hours. gati The IS demos program returns to AAAI-02 for its fourth year. Continuing advances in artifi- 11:00 AM:FlexBot, Groo, Patton and Ghost- cial intelligence research are making it possible Bot: Research using Computer Games as a to develop intelligent systems in a wide range Platform (Booth D316) Aaron Khoo, Robin of application areas. The AAAI-02 Intelligent Hunicke, Greg Dunham, Nick Trienens, and Systems Demonstrations program showcases Muon Van state-of-the-art AI implementations and pro- 1:00 PM:Fuzzy Neural Networks in a Palm En- vides AI researchers with an opportunity to vironment (Booth D421) Samuel Moyle show their research in action. and Michael Watts The program is intended to highlight inno- 2:00 PM:Research Applications of the MAG- vative contributions to the science of AI with NET Multi-Agent Contracting Testbed an emphasis on the benefits to be gained from (Booth D415) John Collins and Maria Gini developing and using implemented systems in 3:00 PM:AI Festival (all demos available) AI research. Previous year’s demonstrations have included speech- and gesture-based systems, AI-based simulators and game-playing systems, several systems using AI on the world Demo Descriptions wide web for e-commerce and other applications, and even AI pets. System builders will be Booth D320 on hand to present their work, and audience An Automated Negotiator for an interaction with the systems is encouraged as International Crisis much as possible. Penina Hoz-Weiss, Sarit Kraus, Jonathan Wilkenfeld All demonstrations will be available during and Tara E. Santmire, Bar-Ilan University, Ramat the AI Festival on Wednesday afternoon, and Gan, Israel and University of Maryland, College Park the tentative individual demonstration sched- We present an automated agent that can nego- ule for Tuesday and Wednesday is listed below. tiate efficiently with humans in bilateral nego- tiations with time constraints and the possi- Demonstrations Schedule bility of opting out. The negotiation is con- ducted using a semiformal language. The Tuesday, July 30 model used in constructing the agent is based on a formal analysis of the scenario using 10:00 AM:JYAG & IDEY: A Template-Based game theoretic methods and heuristics for ar- Generator and Its Authoring Tool (Booth gumentation. The agent receives messages sent D314). Songsak Channarukul, Susan W. by humans, analyzes them and responds. It al- McRoy, and Syed S. Ali so initiates discussion on one or more param- 11:00 AM:Constructive Adaptive User Inter- eters of an agreement. The specific scenario faces – Composing Music Based on Human concerns a crisis between Spain and Canada Feelings (Booth D417). Masayuki Numao, over access to a fishery. Shoichi Takagi, and Keisuke Nakamura The human players are provided with a 1:00 PM:An Automated Negotiator for an In- DSS to analyze the scenario and compare the ternational Crisis (Booth D320). Penina utility points associated with various out- Hoz-Weiss, Sarit Kraus, Jonathan Wilkenfeld comes, and with a language editor to facilitate and Tara E. Santmire the composition of messages. In the demo, we 2:00 PM:SpeechWeb: A Web of Natural-Lan- will present the support tools for the humans guage Speech Applications (Booth D419). and then demonstrate a negotiation session Richard Frost between the agent and a human. Audience members will have an opportunity to negoti- 3:00 PM:UTTSExam: A University Examina- ate with the agent. tion Timetable Scheduler (Booth D312). Andrew Lim, Juay-Chin Ang, Wee-Kit Ho, and Wee-Chong Oon 21 Booth D417 Booth D316 Constructive Adaptive User Inter- FlexBot, Groo, Patton and GhostBot: faces – Composing Music Based on Research using Computer Games as a Human Feelings Platform Masayuki Numao, Shoichi Takagi, and Keisuke Aaron Khoo, Robin Hunicke, Greg Dunham, Nick Nakamura, Department of Computer Science, Trienens, and Muon Van, Northwestern University Tokyo Institute of Technology This is a demonstration of FlexBot, a research We demonstrate a method to locate relations platform built using the Half-Life game en- and constraints between a music score and its gine, and three systems based on the FlexBot impressions, by which we show that machine architecture. Groo, our resident Half-Life bot, learning techniques may provide a powerful uses behavior-based techniques to make criti- tool for composing music and analyzing hu- cal decisions in real time. Patton is a recently man feelings. The demonstration introduces constructed system for monitoring and con- two user interfaces, which are capable of pre- trolling bot teams in Half-Life using remote dicting feelings and creating new objects based devices such as PDAs. GhostBot is part of a sys- on seed structures, such as spectrums and their tem designed to monitor a player’s behavior in transition for sounds that have been extracted a game and adjust the environment to facilitate and are perceived as favorable by the test sub- a smooth, yet challenging play experience. ject. The interfaces first collect a person’s feel- The demonstration is designed to show ings for some pieces, based on which they ex- FlexBot in action and to exhibit the flexibility, tract a common musical structure causing a efficiency and overall ease with which the specific feeling. One arranges an existing song FlexBot architecture supports a variety of AI to fit such a structure causing a specified feel- research tasks. During the demo, conference ing. Another composes a new piece. attendees will be able to observe FlexBot agents in action, as well as play games against them. Booth D413 Disciple-RKF/COG: Agent Teaching Booth D421 by Subject Matter Fuzzy Neural Networks in a Gheorghe Tecuci, Learning Agents Lab, George Mason Palm Environment University Samuel Moyle and Michael Watts, Department of Disciple-RKF/COG is a learning agent shell Information Science, University of Otago that can perform many knowledge engineering tasks, and can be used to develop knowledge Booth D314 based systems by subject matter experts, with JYAG & IDEY: A Template-Based limited assistance from knowledge engineers. Generator and Its Authoring Tool The expert and the agent engage into a mixed- initiative reasoning process during which the Songsak Channarukul, Susan W. McRoy, and Syed S. Ali, University of Wisconsin-Milwaukee expert is teaching the agent his problem solving JYAG is the Java implementation of a real- expertise, and the agent learns from the expert, Intelligent Demos Systems time, general-purpose, template-based gener- building, verifying, and improving its knowl- ation system (YAG, Yet Another Generator). edge base. JYAG enables interactive applications to adapt The demonstration will show how a trained natural language output to the interactive Disciple agent helps the students at the US context without requiring developers to write Army War College to learn about Center of all possible output strings ahead of time or to Gravity analysis. Then, the main part of the de- embed extensive knowledge of the grammar mo will show how this Disciple agent was of the target language in the application. Cur- taught the problem solving expertise of a mili- rently, designers of interactive systems who tary expert. The demonstration will conclude might wish to include dynamically generated with a presentation of several ontology import text face a number of barriers; for example de- and development tools of Disciple-RKF/COG signers must decide (1) How hard will it be to that are used by a knowledge engineer to perlink the application to the generator? (2) Will form knowledge base development tasks that the generator be fast enough? (3) How much are currently beyond the capabilities of a sub- linguistic information will the application ject matter expert. need to provide in order to get reasonable quality output? (4) How much effort will be required to write a generation grammar that covers all the potential outputs of the applica- 22 tion? The design and implementation of our decision processes. The other shows how the Systems Demos Intelligent template-based generation system, JYAG, is in- open-source MAGNET system can be config- tended to address each of these concerns. A ured to support experimentation with delibera- template-based approach to text realization re- tion scheduling and winner determination. quires an application developer to define templates to be used at generation time; therefore, Booth D419 the tasks of authoring and testing templates are SpeechWeb: A Web of Natural-Lan- inevitable. JYAG provides predefined tem- guage Speech Applications plates. Developers may also define their tem- Richard Frost, Department of Computer Science, plates to fit the requirements of a domain-spe- University of Windsor, Canada cific application. Those templates might be to- The demonstration shows an enhanced natu- tally new or they can be a variation of existing ral-language speech browser navigating a web templates. Even though developers can author of remote hyperlinked applications. The a template by manually editing its textual defi- browser recognizes relatively complex spoken nition in a text file, it is more convenient and input such as “which moon that was discov- efficient if they can perform such tasks in a ered by Hall does not orbit mars?” and sends graphical, integrated development environ- such input to a remote interpreter that is ac- ment. The integrated development environ- cessed through the Internet. Users can ask to ment for YAG (IDEY) provides these services move from one application to another in a as a tool for JYAG’s templates authoring, test- manner that is analogous to following a hy- ing, and managing. IDEY’s graphical interface perlink on a regular web page. The browser reduces the amount of time needed for syntax runs on a regular PC and uses off-the-shelf familiarization through direct manipulation IBM ViaVoice speech-recognition software. and template visualization. It allows a develop- The input is spoken through a wireless micro- er to test newly constructed templates easily. phone giving hands-free, eyes-free, access to remote data sources. The demonstration illus- Booth D318 trates application of a new semantics for nat- Multi-ViewPoint Clustering ural-language processing that accommodates Analysis Tool arbitrarily-nested quantification and nega- Mala Mehrotra Pragati, Synergetic Research Inc. tion, and also a new technique for improving The multi-viewpoint clustering analysis (MVP- speech-recognition accuracy. CA) technology uses clustering analysis techniques to group rules of a knowledge base that Booth D312 share significant common properties. We dem- UTTSExam: A University onstrate a research prototype tool that enables Examination Timetable Scheduler knowledge engineers and subject matter experts Andrew Lim, Juay-Chin Ang, Wee-Kit Ho, Wee- (SMEs) to familiarize themselves rapidly with Chong Oon, School of Computing, National Uni- the terms and concepts in a knowledge base, to versity of Singapore exploit and reuse preexisting knowledge, and to UTTSExam is a university examination merge and align concepts across different timetable-scheduling program, customized for knowledge bases reliably and efficiently. the National University of Singapore (NUS). The display comprises informational posters Booth D415 and two notebooks running the Registrar and Research Applications of the MAG- Faculty versions of UTTSExam respectively. NET Multi-Agent Contracting Testbed The demonstration begins with an expla- nation of the NUS timetabling problem, in- John Collins and Maria Gini, University of Min- nesota cluding: information on NUS; statistics of the MAGNET is a testbed for exploring decision actual data; scheduling strategy and the rea- processes and agent interactions in the domain soning behind it; the combined method of multi-agent contracting. Experimental re- scheduling algorithm. This is followed by a search in this area requires a simulation envi- demonstration of the software and scheduling ronment that is sufficiently rich to be easily process, including: scheduling of the faculty adapted to a variety of experimental purposes, timetables; merging of the faculty timetables; while being sufficiently straightforward to sup- conflict resolution. The demonstration systems port clear conclusions. Two different demon- will be loaded with actual data from semester 1 strations will be available. One uses a user inter- of the 2001/2002 academic year in NUS. face to help visualize the domain and the agent’s 23 Robot Competition ■ Robot Exhibition: Exhibit Hall AB 11:55 AM - 12:45 PM & Exhibition ■ AAAI Invited Panel on the Robot The Eleventh Robot Competition and Exhibi- Competition and Exhibition tion will be held in Exhibit Hall AB on the as- 3:00 PM - 5:30 PM sembly level of the Shaw Conference Centre, ■ Robot Host: Robot Server, during AI and will be open to registered conference at- Festival tendees during exhibit hours. This series of 5:00 PM - 5:30 PM events brings together teams from colleges, ■ Awards Ceremony: Exhibit Hall AB, universities and other research laboratories to during AI Festival compete, and also to demonstrate state-of- Thursday, August 1 the-art research in robotics and AI. The goals of the Competition and Exhibition are to (1) 9:00 AM – 5:00 PM ■ foster the sharing of research ideas and tech- Robot Workshop (by invitation only), nology; (2) allow research groups to showcase Salon 5, Meeting Level their achievements; (3) encourage students to

& Schedule enter the ﬁelds of robotics and AI; and (4) in-

crease awareness of the field. Robot Competition and Exhibition Teams Competition The competition allows teams to show off Joint Challenge Competitors their best attempts at solving common tasks in Carnegie Mellon University, Naval a competitive environment. Teams compete Research Laboratory, Metrica, North- for place awards as well as for technical inno- western University, and Swarthmore vation awards, which reward particularly in- College teresting solutions to problems. There will be three contest events this year: Robot Host, Robot: GRACE CMU Team: Reid Simmons, Greg Armstrong, Alli- Robot Rescue, and the Robot Challenge. son Bruce, Dani Goldberg, Adam Goode, Illah Nourbakhsh, Nicholas Roy, Brennan Sellner, David Exhibition Silver, Chris Urmson NRL Team: Alan Schultz, Myriam Abramson, The exhibition gives researchers an opportu- William Adams, Amin Atrash, Magda Bugajska, nity to demonstrate state-of-the-art research Mike Coblenz, Dennis Perzanowski Metrica Team: David Kortenkamp, Bryn Wolfe in a less structured environment. Exhibits are Northwestern Team: Ian Horswill, Robert Zubek scheduled throughout the exhibition hall Swarthmore Team: Bruce Maxwell hours. GRACE (Graduate Robot Attending Confer- encE) is a multi-institutional, cooperative ef- Workshop fort consisting of Carnegie Mellon University, the Naval Research Laboratory, Metrica, The robot events culminate with a workshop Northwestern University, and Swarthmore where participants describe the research be- College. This year’s goal is to integrate soft- Robot Programs hind their entries. ware from the various institutions onto a common hardware platform and attempt to Schedule do the complete AAAI Robot Challenge task autonomously, from beginning to end. The fo- Tuesday, July 30 cus will be on multi-modal human robot in- 10:00 AM – 5:30 PM teraction (speech and gesture), human-robot ■ Robot Challenge: Throughout Shaw Con- social interaction, task-level control in the face ference Centre of a dynamic and uncertain environment, ■ Robot Exhibition: Exhibit Hall AB map-based navigation, and vision-based in- ■ Robot Rescue: Exhibit Hall AB teraction. ■ Robot Host: Information Kiosk, during Interacting naturally with humans, GRACE breaks outside Exhibit Hall AB will find its way from the convention entrance to the registration area. It will query by- Wednesday, July 31 standers for directions to the registration desk 10:00 AM – 5:30 PM and navigate there based on those directions. ■ Robot Rescue: Exhibit Hall AB Along the way, it will interact with other con- 24 ferees and will ride in the elevator, using an based on a simple modular diff-drive plat- Teams Robot electronic altimeter to determine when it is on form, mounted with a heterogeneous array of the right floor. It will use color vision to find sensors. Sensor types include both vision and the registration sign, and use laser and stereo pyroelectric sensing. Successful human detec- vision to queue itself and wait in line. It will tion is communicated through two channels: a interact with the volunteer at the registration low-bandwidth channel to ward off neighbor- booth, and use map-based navigation to trav- ing robots; and a coded radio broadcast, indi- el to the Exhibition Hall. Finally, it will present cating success and believed relative location. a talk about itself (see Program addendum for This signal is received by a rescue workstation. time and place). Individual robots accomplish their tasks autonomously, using distinct search strategies. Exhibitor Collective behavior is observed through sim- Carnegie Mellon University ple success-based interactions. (Robotics Institute) Tartan Swarm is a simple, low-cost, educa- tional project of the undergraduate Carnegie Robot: The Personal Rover Project Team Leader: Illah Nourbakhsh Mellon Robotics Club Team Members: Emily Falcone, Rachel Gockley, Eric Porter, Illah Nourbakhsh Exhibitor The Personal Rover Project, funded by Columbia University NASA/Ames Autonomy programs, aims to de- Robot: RoboCupJunior velop an affordable, highly competent mobile Team Leader: Elizabeth Sklar rover that will serve as a scientific and creative Team Member: Simon Parsons outlet for children. This rover is novel in its RoboCupJunior is a project-oriented educa- use of leading-edge microprocessor technolo- tional initiative that sponsors local, regional gy to enable very high robot competence at and international robotic events for students. very low price point. An on-board CMUcam This year marks the third year of internation- vision system tracks colorful objects using just al competition, with a tournament being held a Ubicom microprocessor and a CMOS imag- in conjunction with RoboCup 2002 – ing sensor. A network of on-board PIC micro- RoboCupJunior 2002 will include over 60 processors, called Cerebellum, provide sensor teams of high school and middle school stu- interfaces as well as fine-grained motor con- dents from more than a dozen countries trol using back-EMF speed sensing. An on- world-wide. Teams build and program au- board StrongARM processor provides 802.11b tonomous mobile robots to play soccer, per- networking as well as time-critical visual-mo- form dances and simulate rescue scenarios. tor feedback loops. Finally, using a movable We have also used the RoboCupJunior mo- Center of Mass mechanism, the Personal tif as the theme for undergraduate classes in Rover is able to traverse steps greatly exceed- AI, robotics and programming. The soccer ing its wheel diameter. and rescue contests have been extremely moti- During AAAI we plan to demonstrate both vating and challenging for college students the step traversal and vision-based navigation with a variety of backgrounds and skill levels. competencies of the Personal Rover. In addi- Our exhibition will introduce RoboCupJu- tion to achieving these technological compe- nior to the AAAI audience, in search of men- tencies, the Personal Rover Project also focus- tors for teams of young students as well as ed- es on interaction design in order to produce a ucators looking for a new twist on standard compelling human-robot interface through a undergraduate curriculum. series of formative evaluations. We will also demonstrate such interaction interfaces, Exhibitor painting a picture of how the Rover can be a Dartmouth University tool for conducting simple science experi- Robot: Crystal Robots ments in and around the home. Team Leaders: Robert Fitch and Daniela Rus A robot designed for a single purpose can per- Rescue Competitor form some specific task very well, but it will Carnegie Mellon University perform poorly on a different task, in a differ- (Robotics Club) ent environment. This is acceptable if the en- Robot: Tartan Swarm vironment is structured; however if the task is Tartan Swarm is a low-cost multirobot ap- in an unknown environment, then a robot proach to human detection. Each robot is with the ability to change its shape to suit the environment and the required functionality 25 will be more likely to succeed than a fixed-ar- for the operator and makes different demands chitecture robot. We wish to create more ver- on the robot and communication infrastruc- satile robots by using self-reconfiguration: ture. To meet this objective we are working to- hundreds of small modules will autonomous- wards the following technical goals: (1) sliding ly organize and reorganize as geometric struc- autonomy to support changing communica- tures to best fit the terrain on which the robot tion, cognitive, perceptual and action capabil- has to move, the shape of the object the robot ities of the user and robot; (2) self-monitoring has to manipulate, or the sensing needs for the and continuous assessment of perception and given task. For example, the robot could syn- behavior performance; (3) novel interfaces for thesize a snake shape to travel through a nar- robust, mixed-initiative interaction between row tunnel, and then morph into a six-legged robots and humans; (4) simultaneous local- insect to navigate on rough terrain upon exit. ization and mapping techniques that capture Self-reconfiguring robots are well-suited for an abstracted representation of the robot’s ex- tasks in hazardous and remote environments, perience vis-a-vis the task and environment; especially when the environmental model and and (5) sensor-suites and fusion algorithms to the task specifications are uncertain. A collec- enhance perception capabilities for sensing, tion of simple, modular robots endowed with interpreting, and “understanding” environ- self-reconfiguration capabilities can conform mental features and humans. to the shape of the terrain for locomotion by Through these technical efforts, we will en-

Robot Teams implementing “water-flow” like locomotion able remote robotic operations to be accom- gaits, which allow the robots to move by con- plished by fewer operators with less training. forming to the shape of the terrain. We have This work will pave the way for a new class of designed and built the Crystal robot—which is mixed-initiative robots that work with hu- capable of autonomous shape changing, loco- mans as well as for them, accepting high-level motion by self-reconfiguration, and self-repli- tasks with minimal demands on the user. cation of a big robot into smaller robots—with the same functionality. Challenge Competitor iRobot Corporation Rescue Competitor Robot: CoWorker Georgia Technical University Team Members: Mark Dockser and Jim Allard Robots: Georgia Tech Yellow Jackets The CoWorker robot is an internet controlled, Team Leader: Tucker Balch wireless, mobile, remote telepresence plat- Georgia Tech will compete in the Robot Rescue form. CoWorker can be accessed from any PC competition using a cooperative multi-robot web browser with a high-speed connection system. The robots include an RWI ATRV Mini (and secure ID and password). The user- equipped with 8 DV cameras arranged to pro- friendly interface provides control over where vide omnidirectional vision, and four Sony AI- the CoWorker goes, what it sees, what it hears BO legged robots. The AIBOs are transported and provides an interface for speaking. There by the ATRV, then released to explore areas is even a laser pointer so that the user can where the ATRV cannot reach. The ATRV will highlight what she is referring to at the robot’s provide 3D modeling of the environment as location. The platform was designed with well as localization and tracking of the AIBOs. many available ports (power, serial, PCMCIA) for incorporation of additional hardware in- Exhibitor cluding sensors and/or additional cameras. Idaho National Engineering and Our vision for the CoWorker is to meet the Environmental Laboratory needs of industrial users for: (1) remote expert applications; (2) security; and (3) video- Robot: Junior Team Leader: David Bruemmer conferencing anywhere. By deploying Co- INEEL is working to develop robots that can Workers, customers can dramatically reduce adjust their level of autonomy on the fly, lever- travel costs, allow for collaboration anywhere aging their own, intrinsic intelligence to meet in a company’s wireless network (including whatever level of control is handed down the shop floor, cafeteria or any other location), from the user(s). Currently, we have imple- and improve worker safety (by deploying a mented a control architecture with four dif- CoWorker rather than a human to potentially ferent levels of autonomy: teleoperation, safe- hazardous situations). CoWorker robots are mode, shared control and full autonomy. Each currently in beta applications with a number level of control encompasses a different role of Fortune 500 companies. 26 Host Competitor ed search using a team of robots controlled by Teams Robot Kansas State University a single human. The robots are semiau- Robot: Borivoj tonomous and able to share information di- Team Members: Vojtech Derbek, Jan Kraus, Tomas rectly with one another, and with a human via Tichy, David A. Gustafson a commander console program. The robots will perform obstacle avoid- Exhibitor ance, localization, and low-level route plan- Kansas State University ning autonomously. Mapping, target detec- Robots: Kansa, Wichita, Coronado and Pike tion, and goal-directed behavior will be per- Team Leader: Eric Matson formed via coordination between robots and Robot teams have advantages over individual the human commander. robots in accomplishing goals that contain We are currently using three ActivMedia large numbers of tasks. The advantage grows if Pioneer 2-AT robots, each equipped with they have the ability to interchange roles, share sonar and a single color camera. We expect to responsibility and have some redundant capa- eventually integrate other platforms (e.g. bility. An organization which has the ability to iRobot Packbot) and sensors (e.g. laser continuously evaluate capabilities and role as- rangefinder, pyrosensors, and microphones). signments and reorganize to maximize efficiency will naturally operate at higher levels. Rescue Competitor Our research is to create a cooperative New Roads High School robotics (CR) reorganization model to dy- Robots: Morph Dragon, Ringo namically evaluate and reorganize the team in Team Members: The Scarabs, Michael Randall the event of a failure, or suboptimal executing In 1999, a group of high-school and junior- condition. We are currently building a model high students from Los Angeles took on the and system to allow a team of heterogeneous enormous challenge of competing against robots to conduct real-time reorganizations some of the top robotics and AI researchers in while working in a specific task environment. the world in the RoboCup middle (F2000) league. After over two years of hard work, Challenge Competitor funded on a shoestring budget (mostly out- MIT of-pocket), the Scarabs robotic team field- Robots: Leo and Erik tested a color-tracking system at RoboCup Team Leader: John Leonard 2000 in Melbourne, Australia; designed and Our research addresses the problem of concur- built a prototype vehicle and omnidirectional rent mapping and localization (CML) for au- vision system; and successfully demonstrated tonomous mobile robots. The problem of CML this vehicle-vision system combination in the is stated as follows: starting from an initial po- AAAI/RoboCup Rescue Robot competition in sition, a mobile robot travels through a se- Seattle, Washington. quence of positions and obtains a set of sensor The goals of the Scarabs team are to build measurements at each position. The goal is for viable robots at minimal cost; to learn about the mobile robot to process the sensor data to math, computer science, electronic engineer- produce an estimate of its position while con- ing, physics, AI, system integration, interna- currently building a map of the environment. tional relations, character development, and While the problem of CML is deceptively easy teamwork; to have fun (!); and to make a posi- to state, it presents many theoretical challenges. tive difference. The problem is also of great practical impor- We are fielding two radically different tance; if a robust, general-purpose solution to robots: Ringo, an updated version of the pro- CML can be found, then many new applica- totype we ran in 2001; and Morph-Dragon, a tions of mobile robotics will become possible. sophisticated six-wheeled robot designed to During the robot challenge, we will attempt a compete on the Robotica television program. demonstration of an algorithm for real-time Both robots will use the same vision and con- large-scale CML using multiple submaps trol systems. We have upgraded our custom- built omnidirectional vision system with the Rescue Competitor Axis 2120 Network Camera (www.axis.com). The MITRE Corporation The 2120 features direct connection with a Robots: Moe, Larry, and Curly 10/100 MBit Ethernet network and a built-in Team Members: Zach Eyler-Walker and David Smith Linux web server. This allows a single Ethernet We are developing an approach to coordinat- cable for video and robot control.

27 Exhibitor Host and Rescue Competitor Northwestern University Swarthmore College Robot: Rosey the Robot Robots: Frodo & Gollum Team Leader: Christopher Dac Le Team Leader: Bruce Maxwell The RObot Self-Explains whY (ROSEY) sys- For the Host Competition, we hope to build tem attempts to demonstrate a behavior-based some interesting human-robot interactions robots ability to generate verbal explanations upon our name-tag reading system that we in response to questions about its behavior. developed for the 2001 competition. We also Speciﬁcally, the robot recognizes a class of why hope to have two moving robots this year so questions that seek reasons for its locomotive that they can converse when they see one an- behavior such as, “Why are you turning?” other. Our overall goal, however, will be to ROSEY the Robot will be running around the successfully serve hors d’oeuvres in an unob- exhibit hall while fielding such questions, trusive and effective manner. which will be typed in. In the Rescue Competition we will be ex- ROSEY the Robot represents an instance of amining issues in semi-autonomous robot a class of what we call self-explanatory robots, systems. Our goal is to permit a single user to which should be able to explain what they’re successfully manage more than one robot. We doing and why they’re doing it. To build such will build upon our system from a year ago robots, we are exploring how a behavior-based that identiﬁed victims using vision, built maps

Robot Teams robot can generate explanations by accessing with paths to found victims, and enabled both its own internal structures and relating them teleoperation and completely autonomous to its sensory-motor state. robot functioning.

Rescue Competitor Rescue Competitor Sharif University of Technology Temasek Polytechnic Robots: EMDAD II, Twins and Magellan Pro Robots: Temasek Search and Rescue Robot Team Leader: Dr. Amir Hossein Jahangir Team Leader: Nadir Ould-Khessal Team Members: S. Bayat Sarmadi, G. Asadi, S. Shar- Civil Defence Search and Rescue robot is an iﬁ Tehrani, E. Mozaffaripour, M. Sonee, A. Farahani, attempt to replace human rescue personnel H. Khaleghi, N. Honarmand with tank-like vehicle capable of searching for EMDAD II, Twins and Magellan Pro are the victims in an unstructured terrain. The di- robots used in the competition. These agents mension of the robot is 502 mm in width, 760 are supervised by a central system. Each of the mm in length and 350 mm in height. The esti- robots searches alone, and when it encounters mated weight of the vehicle is 75kg. The front- a suspected object, sends an alarm or a mes- mounted camera can be rotated up to 360 de- sage to the central system. In this case, degree. Civil Defence Search and Rescue Robot is pending on the received information, the cen- equipped with a Charged Coupled Device tral system will realize whether a victim is pre- (CCD) camera, 24 ultrasonic sonar sensors, sent or not, and the process will be kept on till wireless RF receiver, wireless audio receiver, the end of allowed period. and wireless video transmitter, as well as an Yellow Arena: A commercial robot, Magel- onboard processing unit lan Pro is used in this arena. The main challenge has been to develop software capable of Exhibitor motion planning and visual navigation. Orange Arena: Two robots, Twins, are con- Universita di Roma “la Sapienza” sidered for this arena. The mechanical four- Robot: ArmHandOne wheel platform is based on a mobile toy struc- Team Members: M. Cialente, A. Finzi, I. Mentuccia, F. Pirri, M. Pirrone, M. Romano, F. Savelli, K. Vona ture. An industrial small size PC does the in- We have a reconﬁgurable maze—that is, a formation processing. Twins use ultrasonic maze made of panels that can be suitably ar- sensors, IR sensors and IR cameras. ranged. Inside this maze we can position sev- Red Arena: A special robot has been de- eral road signs indicating whether the road is signed for this section namely EMDAD II. The one-way or no-entry, arrows addressing direc- hardware platform of EMDAD II, is a reliable tions for the exit, and so on. We are free to put system based on industrial PC104 boards. the road signs inside the maze, that is, the This hardware system is capable of operating agent performance should not depend on the in hostile environments. signs. Furthermore there is a place in the maze in which we can locate a treasure, consisting of

28 a set of colored blocks, suitably arranged: e.g. the Lazzo of the Statue, in which Arlecchino Teams Robot forming towers etc. The task for the agent is to pretends to be a statue who moves when the get into the maze, find a particular block (e.g. backs of the other actors are turned. the red block), picking it up (to achieve this Our troupe of three robots perform a short the robot might need to move lots of other script, with each executing its own plan, cue- blocks) and direct itself toward the exit and, fi- ing off the others to keep themselves in sync. A nally, exit from the maze. human director observing the performance The robot we are talking about is named can affect the overall performance, or the per- Armandone (pronouced “ArmHandOne”), formance of an individual robot. The robots weighing about 4 kg and standing about 40 cm themselves consist of Lego bases, carrying a high. It is equipped with a grabber arm, a pan- HandyBoard processor, a two-way radio link, tilt binocular head with two cameras, and oth- and the puppet body. er sensors. Wireless control can reach 1 km. The novelty of our approach mainly relies in Rescue Competitor the cognitive architecture we have been build- University of Manitoba, Winnipeg, ing. The architecture is defined on three levels: Canada (1) the cognitive level monitoring sensing and Robots: Keystone Fire Brigade high level control actions, (2) the global level Team Leader: Jacky Baltes managing the choice of tasks and actions con- Team Members: Jacky Baltes, John Anderson trol, and (3) the reactive level managing navi- The Keystone Fire Brigade robots are based on gation and localization. the 4 Stooges, a small sized RoboCup team from the University of Auckland. The robots Exhibitor, Challenge and Host Competitor of the 4 Stooges were designed to be robust Universidade de Aveiro, Portugal and versatile enough to be used in a variety of Robot: Carl different ways. This has paid off since the Team Leader: Luis Seabra Lopes robots of the Keystone Fire Brigade are identi- Team Members: Luis Seabra Lopes, Antonio Teix- cal to those. eira, Mario Rodrigues, Qinghua Wang and three un- The Keystone Fire Brigade use a small dergraduate students CMOS camera and Thomas Braunl’s Eyebot CARL is an acronym for “communication, ac- controller. The Eyebot controller consists of a tion, reasoning and learning in robotics.” The 35 MHz 68332 processor with 2 MB of static project is therefore concerned with the devel- RAM. The design is clearly dated nowadays, opment of an integrated set of capabilities for but has the advantage that it is comparatively robot intelligence. The participation will build cheap and provides the possibility of directly upon previously implemented capabilities connecting a CMOS camera to the processor (multisensor navigation, spoken-language di- as well as providing the necessary interface to alog), already demonstrated at the 2001 AAAI connect motors, servos, gyroscopes, and many Competition. A higher degree of interactivity other sensors directly to the controller. will hopefully be demonstrated. For the Challenge competition, Carl will Exhibitor perform only a part of the proposed task. Ca- pabilities developed for the participation of University of Minnesota Carl at the Host competition will also be used Robot: MinDART (Minnesota Distributed Au- here. In addition, we may demonstrate on-line tonomous Robotic System) Team Leader: Paul E. Rybski learning. Team Members: Paul E. Rybski, Amy Larson, Chris Flowers-Huebner, Maria Gini Exhibitor The Minnesota Distributed Autonomous University of Connecticut Robot Team (MinDART) is a group of simple Robots: I Comici Roboti and low-cost robots used at the University of Team Leader: Karl R. Wurst Minnesota for research into reactive control Combining Robotics, Puppetry, and Comedy, strategies. We are interested in studying how our troupe of three robots perform a lazzo environmental and control factors affect the from the Commedia Dell’Arte. The Italian performance of a homogeneous multi-robot Comedies of the 16th and 17th centuries had team doing a search and retrieval task. Several many improvisational pieces called lazzi. factors which affect the performance of the These were comic interludes inserted by a team are examined. One factor is the distribu- player if a scene started to drag or his elo- tion of targets. This is varied from a uniform quence gave out. I Comici Roboti performs distribution to having all of the targets clus- 29 tered together into one or two small clumps. Exhibitor Another factor that we examine is the size of USC/Information Sciences Institute the team (varying from one to five robots). Fi- Robot: CONRO nally, the type of search strategy is varied be- Team Leaders: Wei-Min Shen and Behnam Salemi tween a completely reactive method to a direct- The CONRO Project has a goal of providing ed search method that uses the robot’s ability to the Warfighter with a miniature reconfig- localize itself. Current work includes incorpo- urable robot that can be tasked to perform re- rating the ability for the LEGO robots to com- connaissance and search and identification municate amongst themselves using an RF data tasks in urban, seashore and other field envi- link and determining under what environmen- ronments. CONRO will be miniature and is to tal conditions such communications is useful. be made from identical modules that can be programmed to alter its topology in order to Exhibitor, Host, and Rescue Competitor respond to environmental challenges such as University of Rochester obstacles. The base topology is simply con- Robot: Mabel the Mobile Table nected, as in a snake, but the system can re- Team Members: Undergraduate Robot Research configure itself in order to grow a set of legs or Team other specialized appendages. Each module We have developed a Java application which will consist of a CPU, some memory, a battery, we call the “Learn Server.” Its purpose is to and a micro-motor plus a variety of other sen-

Robot Teams provide offboard graphical user interfaces and sors and functionality, including vision and parameter adjustment modules to offboard wireless connection and docking sensors. Ma- robotics applications. It allows communica- jor challenges include packaging, power, and tion with programs running in multiple lan- cooling, as well as the major issue of program- guages and on multiple operating systems. ming and program control. Currently the program communicates with the C and C++ languages, in Windows2000, Exhibitor WindowsXP, and Linux. We will be demon- University of Texas at Dallas strating this application in the context of our Robot: Identity Emulation (IE), Facial Expression robot host system: Mabel the Mobile Table. Robot Mabel gives an appropriate multi-modal Robot Team Members: David Hanson, Marshall response to people using a combination of Thompson, Giovanni Pioggia speech, food manipulation, and navigation Our facial expression robot uses biomimetic behaviors. We accomplish this using the structures, aesthetic design principles, and re- Sphinx Speech Recognition System developed cent breakthroughs in elastomer material sci- by CMU, augmented by a digital filter. We em- ences to enact a sizable range of natural hu- ploy a directed speech recognition micro- manlike facial expressions. This application of phone, which is actively pointed towards the robotics will rise in relevance as humans and speaker, using face tracking and a pan-tilt- robots begin to have more face-to-face en- zoom camera. To accomplish language under- counters in the coming years. My team and I standing, a specially designed grammar-based are also working on imbuing our robot with parsing technique is under development. several forms of interactive intelligence, in- The vision component’s purpose is to pro- cluding human-form and facial-expression vide the navigation component with real-time recognition, and natural language interaction. visual percepts including estimated closest pa- It is anticipated that an integration of me- tron group, best approach angle in field of chanics, sociable intelligence, and design aes- view, patron face-region detection and track- thetics will yield the most effective human ing, nametag-region detection scene, and computer interface robots. character classification (ANN) Autonomous navigation control involves Rescue Competitor creating a robust model for navigating Utah State University around a crowded room while retaining the Robots: Blue Swarm 2 and Blue Swarm Sentinel ability to return to a base station. We use Team Members: Asti Bhatt, Brandon Boldt, Scott Sk- sonar-based obstacle avoidance for robust ousen, and Dan Stormont navigation. To achieve path planning and ex- The Blue Swarm 2 is made up of six modified ecution, we employ a trained waypoint sys- remote-control cars. They operate au- tem using wheel counters. tonomously using a simple subsumption architecture. They will sense the location of victims and send out a signal which can be re- 30 Botball ceived by one or more Blue Swarm Sentinels. The Blue Swarm Sentinel is a modified radio- National Botball Exhibition controlled tank that operates either manually No, the graduate students haven’t gotten or autonomously to locate victims, locate Blue younger! AAAI is pleased to host the National Swarm robots that have located a victim, or Botball Exhibition, featuring top robots built locate obstacles. The Sentinel reports infor- by middle and high school students from mation back to the rescuer GUI via a bi-direc- across the country. Botball is a game in which tional RF link. Both types of robots are con- robots attempt to achieve a specified goal, in trolled by Parallax BASIC Stamps. an exciting head to head, double elimination tournament. The goal of Botball is to get mid- Exhibitor dle and high school students involved in the Utah State University creative side of technology - to get our up- Robot: Blue Swarm 3 coming workforce excited about technology, Team Leader: Dan Stormont robotics, and AI. Botball involves embodied Blue Swarm 3 is the next generation of Blue agent computer programming (in C), me- Swarm robots. The Blue Swarm 3 will be built chanical design, science, math, and teamwork. to compete in the Urban Search and Rescue In this year’s tournament, teams either play competition in 2003. The current plan is to the black ball or white ball side. The challenge build robust, legged robots that communicate is to score points by moving your colored ping with each other and with a handheld terminal pong balls from inside a moveable goal into (PalmPilot) using IR links. The exhibit will the basket or into the end-zone. Robots are re- demonstrate some prototypes of the robots quired to start by themselves and shut down that will be developed for Blue Swarm 3. after 90 seconds. These robots were completely designed, Rescue Competitor built, and programmed by students from a kit YSC (Iran) of over 2000 parts. Students first compete in Robot: Hanif 1, Hanif 2 (Snake Robot), Hanif 3 one of 12 regional tournaments and then ad- (Tracked Robot) vance to the National Botball Tournament, Team Leader: Navid Ghaffarzadegan Team Members: Mohammad Hossein Taghavi, Amir held in Norman, Oklahoma in early July. The Salman Avestimehr, Sadegh Dabiri, Arezu Moteval- best of the best from that tournament will be izadeh, Hadi Emamifar, Ali Sharifi, Peiman Shariat showcasing their robots at AAAI this year. Panahy, Sareh Ebrahimi, Mohammad Hossein Far- For more information about the Botball pro- dad, Ali Azimi, Sina Rastad, Mohammad Na- gram, please see www.botball.org. jmzadeh Because of the geographical placement of Event Schedule Iran, every year earthquakes are both com- Tuesday, July 30 mon and fatal. Thus, working on robots that 10:00 AM - 12:00 PM:Botball Practice Rounds could help rescuers in detecting victims is a 1:00 PM - 4:00 PM:Botball Mini-Tournament necessary task. Our aim in this project is to 4:00 PM:Team Programming Challenge design, construct, and control an autonomous Wednesday, July 31 robot that could move around in an unstruc- 10:00 AM - 12:00 PM:Botball Practice Rounds tured environment and detect victims in haz- with new programming ard areas. Our team is divided into three sub- 1:00 PM - 3:00 PM:Botball Seeding Rounds teams: the mechanical engineering students 3:00 PM - 5:30 PM:Botball Mini-Tournament group, the software engineering students (Botball Students & RBL Participants) group, and the hardware engineering students group. Current research includes dynamical analysis of mobile plates; path planning and victim detection using image processing; posi- tioning of mobile robots; design, fabrication, and control of snake robots; design, fabrication, and control of rovs; and mechanical use of smart materials—especially SMA.

31 Registration

Conference registration is located on the assembly level of the Shaw Conference Centre, beginning Sunday, July 28. Registration hours are: Sunday, July 28 7:30 AM – 6:00 PM Monday, July 29 7:30 AM – 6:00 PM Tuesday, July 30 7:30 AM – 5:30 PM Wednesday, July 31 8:00 AM – 5:30 PM Thursday, August 1 8:00 AM – 2:00 PM Only checks drawn on Canadian banks, VISA, MasterCard, American Express, government purchase orders, traveler’s checks, and Canadian currency will be accepted. We cannot accept foreign currency or checks drawn on foreign banks.

Registration Fees The AAAI-02/IAAI-02 technical program registration fee includes admission to the technical plenary and poster sessions, the Exhibition Program, the Tutorial Forum, the Workshop Pro- gram (by invitation only), the Opening Reception, the AI Festival, and the AAAI-02/IAAI-02 Registration Conference Proceedings. Tutorial Forum attendees may register for up to four consecutive tutorials, and will receive the corresponding syllabi. Students must present proof of full-time student status to qualify for the student rate. Onsite technical program fees are in Canadian Dol- lars and are as follows: Regular Member: $1,070 CDN Regular Nonmember: $1,225 CDN Student Member: $345 CDN Student Nonmember: $455 CDN

Workshop Program Workshop registration is limited to those active participants determined by the organizer pri- or to the conference. All workshop participants must register for the AAAI-02 technical program. Registration onsite for a workshop is possible with the prior permission of the corresponding workshop organizer.

Robot Building Lab The robot building lab registration includes admission to the robot building lab and the exhibition program. Fees are $230.00 CDN for members or nonmembers, and $115.00 CDN for students. Attendance is limited and preregistration is required.

Exhibition Admission to the exhibition hall programs is included in all other types of registration. For individuals interested in admittance to the exhibit hall only, an exhibits only registration is available in onsite registration. Exhibit hall programs include vendor exhibits, the Intelligent Sys- tems Demonstrations, the Botball Exhibition and the Robot Competition and Exhibition. High-school students are welcome and will be admitted without fee upon presentation of a valid high-school student ID. Children under 12 will also be admitted without fee, but must be accompanied by an adult conference registrant. Please note: The AI Festival, which will be held in the exhibit hall, is included in the technical registration fee only. All other attendees must pay an additional fee. All passes are good for Tuesday and Wednesday, July 30-31 Exhibits Only Adult: $15.00 CDN Exhibits Only Adult w/Festival: $38.00 CDN Exhibits Only Child w/Festival: $8.00 CDN

32 Admission Staples Business Depot General In Edmonton Core, 10330 – 101 Street Each conference attendee will receive a name Monday – Friday, 8:00 AM - 9:00 PM;Saturday, badge upon registration. This badge is re- 9:00 AM - 6:00 PM;Sunday, 11:00 AM - 5:00 PM quired for admittance to the technical, tutori- 424-4114. Copy center and business services. al, exhibit, IAAI and workshop programs. Workshop attendees will also be checked off a master registration list at individual rooms. Career Information Tutorial attendees must present syllabi tickets A bulletin board for job opportunities in the to receive syllabi. Smoking, drinking and eat- artificial intelligence industry will be made ing are not allowed in any of the technical, tu- available in the registration area, on the assem- torial, workshop or IAAI sessions. bly level of the Shaw Conference Centre. At- tendees are welcome to post job descriptions of Baggage Holding openings at their company or institution. There is no baggage holding area at the Shaw Conference Centre. Please check your luggage Child Care Services formation with the bellman at your hotel after you have For information about childcare services, you checked out. Neither the AAAI, the Shaw Con- may contact Jan Pat Management Program in ference Centre, the Weston Edmonton, the Edmonton at 448-1883. They have babysitting Fairmont Hotel Macdonald, the Crowne Plaza services and KidScenes for families traveling Chateau Lacombe, nor the University of Al- to the city for conferences. (This information berta accept liability for the loss or theft of any is provided for your convenience and does not suitcase, briefcase, or other personal belong- represent an endorsement of this agency by ings brought to the site of AAAI-02/IAAI-02. AAAI. Responsibility for all child care arrangements must be assumed by the parents.) Banking The closest banks and automated teller ma- Coffee Breaks chines (ATM) are below. The ATM networks Coffee will be served in the foyer on the meet- available are Discover, MasterCard, Visa, Cir- ing level of the Shaw Conference Centre, Sun- rus, Honor and Plus. The banks can also ex- day, July 28 and Monday, July 29, & 10:30 – change all major foreign currencies. 11:00 AM and 3:30 – 4:00 PM.Coffee will also Canadian Western Bank be served in the ballroom foyer of the Westin 10303 Jasper Avenue Edmonton during these times. 423-8801 Coffee will be served in the foyer on the as- Monday – Wednesday and Friday, 9:30 AM - 4:30 sembly level of the Shaw Conference Centre PM,Thursday, 9:30 AM - 5:00 PM Tuesday, July 30, 10:30 – 11:15 AM and 3:15 – Royal Bank 3:50 PM,Wednesday, July 31, 10:40 – 11:10 AM, 10107 Jasper Avenue and Thursday, August 1, 10:30 – 11:05 AM and 448-6611 3:00 – 3:20 PM. Monday – Friday, 9:30 AM - 4:00 PM, TD Canada Trust Copy Services 148 Edmonton Centre 448-8000 Copy service is available in the banquet office Monday – Friday, 9:30 AM - 4:00 PM, located outside of Exhibit Hall A. Also see Business Centers. Business Centers Currency The following business centers are available: Canadian dollars will be required for onsite Shaw Conference Centre registration, exhibits only passes, and tutorial Banquet Office located outside of Exhibit Hall syllabi purchases. Approximate rates of ex- A. Services include faxing and photocopying change at press time were: in small quantities. CDN $1.00 = 0.65 US Dollars Westin Edmonton Concourse Level: CDN $1.00 = 0.44 British Pounds CDN $1.00 = 82 Japanese Yen Monday – Friday, 8:00 AM - 5:30 PM CDN $1.00 = 0.69 Euros 414-5700. Copy center and business services. 33 Dining Parking Edmonton dining information is available in The Westin Edmonton charges $16.00 CDN the Edmonton Where Magazine, which has per day for self-parking and $21.00 for valet been included with your registration materi- parking per day. The Crowne Plaza Chateau als. A concession stand (coffee, pastry items, Lacombe charges $7.49 CDN per day for self- candy, etc.) will be open on the assembly level parking and $10.70 CDN for valet parking per during normal business hours, Sunday, July 28 day. The Fairmont Hotel Macdonald charges – Thursday, August 1. The Shaw Conference $8.00 CDN for self-parking or valet for up to 4 Centre is connected via Pedway to Canada hours and $19.95 CDN for more than 4 hours. Place, which has a food court, open 7:30 AM – The City of Edmonton provides over 1,600 4:30 PM,7 days a week. convenient parking stalls within a 5-minute walk from the Shaw Conference Centre. The Handicapped Facilities Library, Canada Place and City Hall Parkades provide pedway connections to the Centre.

formation The Shaw Conference Centre, The Westin Ed- Parking is also available at the City Market and monton, the Fairmont Hotel Macdonald, and on-street meters in the vicinity. the Crowne Plaza Chateau Lacombe are all ■ Canada Place Parkade: 9700 Jasper Avenue. equipped with handicapped facilities. Entrance at 97 Street North of JasperAve. ■ Library Parkade: 10165 – 100 Street. En- Housing trance on 99 Street (Citadel) & 100 Street ■ City Hall Parkade: 1 Sir Winston Churchill For information regarding hotel reservations, Square. Entrance on 99 Street please contact the hotels directly. For student ■ City Market Surface Lot: 10165 – 97 Street. housing, please contact University of Alberta Entrance on 101 and 102 Avenue at 492-4281 or 800-615-4807 (Canada only). Rates: GST included, $1.25 CDN per half hour or part thereof. Information Desk Monday-Friday An information desk/message desk will be 6:00 AM-6:00 PM $10.00 maximum staffed during registration hours, Sunday 6:00 PM-1:00 AM $2.00 maximum Daily Maximum $15.00

General In General through Thursday, July 28 – August 1. It is located near the registration area, on the assem- Saturday and Sunday bly level of the Shaw Conference Centre. Mes- Midnight to 6:00 PM sages will be posted on the message boards ad- ﬁrst 3 hours $1.00 maximum jacent to the desk. The telephone number for Parking thereafter $1.25/half hour leaving messages only is (780) 917-7881. Pag- 6:00 PM-1:00 AM $2.00 maximum ing attendees is not possible. Daily Maximum $10.00

Internet Room Press Internet access is available in Salon 5 of the All members of the media are requested to reg- Shaw Conference Centre. The internet room ister in the Press Room, on the meeting level of will be open Sunday, July 28 - Wednesday, July the Shaw Conference Centre in Salon 20. Press 31, 8:00 AM – 5:30 PM; and Thursday 8:30 AM – badges will only be issued to individuals with 3:30 PM.As a courtesy, please limit your access approved credentials. The Press Room will be time to 10 minutes if others are waiting. open during the following hours. Sunday, July 28 9:00 AM - 5:00 PM List of Attendees Monday, July 29 8:00 AM - 5:00 PM Tuesday, July 30 8:00 AM - 5:00 PM A list of preregistered attendees of the confer- Wednesday, July 31 8:00 AM - 5:00 PM ence will be available for review at the AAAI Thursday, August 1 8:00 AM - 12:00 PM Desk in the registration area on the assembly A volunteer will be on duty during these hours level of the Shaw Conference Centre. Attendee to assist the members of the press and media. lists will not be distributed. Printed Materials Message Center Display tables for the distribution of promo- See Information Desk tional and informational materials of interest 34 to conference attendees will be located in the copies of the conference CD-ROM onsite. General In registration area on the assembly level of the Registrants must pick up their conference Shaw Conference Centre. CD-ROM’s onsite. (AAAI cannot redeem tickets after the conference.) Proceedings Proceedings Shipping Each registrant for the AAAI-02/IAAI-02 technical program will receive a ticket with the Goodkey Show Services at the Shaw Confer- registration materials for one copy of the con- ence Centre can arrange for shipping services. ference proceedings. During registration Contact: Calvin Goodkey at 426-2211. hours on Sunday, July 28, Monday, July 29 and until 10:00 AM on Tuesday, July 30, Proceed- Speaker Ready Room ings tickets can be redeemed at the AAAI Press Proceedings desk, located in AAAI onsite reg- The Speaker Ready Room will be located in Sa- istration on the assembly level of the Shaw lon 19 on the assembly level of the Shaw Con- Conference Centre. After 10:00 AM on Tues- ference Centre. This room has audio-visual formation day, the Proceedings ticket may be redeemed equipment to assist speakers with their prepa- at the MIT Press booth #104, located in Ex- rations. It is important that speakers visit this hibit Hall AB of the Shaw Conference Centre, room to organize their materials. The room will during exhibit hours. Limited Proceedings be open from 8:00 AM - 5:00 PM,Sunday, July 28 will be available at the AAAI Press Proceedings – Tuesday, July 30, and Thursday, August 1; 8:00 desk again on Thursday, August 1 from 8:30 – AM – 3:00 PM on Wednesday, July 31. 1:00 PM.Extra Proceedings may be purchased Invited speakers are asked to come to Salon at the conference site at the above locations. 19 one-day prior to their speech. Representa- Thursday, August 1, will be the last day to pur- tives from AV Headquarters will be available chase extra copies of the Proceedings on site. during the times listed above to conﬁrm your We strongly recommend that you redeem audiovisual needs, and assist with the prepara- your Proceedings ticket onsite.(See “Proceed- tion of your materials, if necessary. ings Shipping” if you do not want to carry your Proceedings with you.) Your ticket can be re- Taxes/Rebate of Taxes deemed by mailing the ticket with your name, complete shipping address and e-mail to: A federal tax (GST) of 7% is now applied to AAAI, Attn: AAAI-02 Proceedings most goods and services in Canada. A harmo- 445 Burgess Drive, Suite 100 nized sales tax (HST) is added to all goods and Menlo Park, California 94025 USA services. Nonresident visitors can apply for a Postage must be prepaid with a check drawn GST/HST rebate on most goods purchased for on a US bank or MasterCard/Visa and expira- use outside Canada as well as on short-term action date. USA: $25.00; for orders outside commodations. Rebates can be claimed in per- USA: $55.00 for regular air mail, $105.00 for son at participating Canadian Duty Free Shops FedEx. Please allow eight to twelve weeks to at border crossings or airports, or by complet- receive your proceedings. ing and mailing the rebate form. A nonresident visitor qualiﬁes for a tax refund if the following conditions are met: you are not a resident of Proceedings on CD-ROM Canada; you purchased eligible goods, short Each registrant for the AAAI-02/IAAI-02 term accommodation or both; you paid technical program will receive a ticket with the GST/HST on these purchases; you have original registration materials for one copy of the con- receipts; the total of your purchase amounts ference CD-ROM. During registration hours (before taxes) for eligible goods and accommo- on Sunday and Monday, July 28-29 and until dations must be at least $200 CDN; each indi- 10:00 AM on Tuesday, July 30, CD-ROM tickets vidual receipt for eligible goods shows a mini- can be redeemed at the AAAI Press Proceed- mum total purchase (before taxes) of $50 CDN; ings desk, located in AAAI onsite registration the goods are removed from Canada within 60 on the assembly level of the Shaw Conference days of the date they were purchased. Retailers Centre. After 10:00 AM on Tuesday, the CD- have forms available. ROM ticket may be redeemed at the Registra- tion Desk. Extra CD-ROM tickets may be pur- Telephones chased in onsite registration. Thursday, Au- gust 1 will be the last day to purchase extra Public telephones for domestic and interna- 35 tional calls are located throughout the Shaw Tutorial Syllabi Conference Centre on all levels. Extra copies of AAAI-02 tutorial syllabi will be available for purchase in AAAI onsite registra- Tipping tion in the Shaw Conference Centre, begin- In general, a tip of 10-20 % is given to waiters, ning Tuesday, July 30. Supplies are limited. waitresses, taxi drivers, etc. Bellhops, doormen, The cost is $25.00 CDN per syllabus. Preregis- porters, etc. at hotels, airports and railway sta- tration tutorial syllabi tickets may be re- tions are generally paid $1.00 per item. deemed in the tutorial rooms.

Transportation Visitor Information The following information provided is the Edmonton Tourism welcomes you to Edmon- best available at press time. Please conﬁrm ton! They are located in the Shaw Conference fares when making reservations. Centre and are open Monday through Friday,

formation 8:00 AM – 4:30 PM. Airlines and Rental Cars Edmonton Tourism The American Association for Artificial Intelli- 9797 Jasper Avenue gence has selected Air Canada as the official Edmonton, Alberta, Canada T5J 1N9 carrier and Avis Rent A Car as the official car Telephone: 426-4715 rental agency for AAAI-02/IAAI-02. If you need to change your airline or car rental reservations, please call Stellar Access, our official travel Volunteer Station agency at 858-805-6109 and reference AAAI The volunteer station will be located in the reg- Event #1889. Web: www.stellaraccess.com istration area on the assembly level of the Shaw Airport Shuttle Conference Centre. All volunteers should check in at the beginning of their shifts with Sky Shuttle is the official carrier to and from the volunteer at the booth prior to their shifts. Edmonton International Airport. They offer The volunteer meeting will be held Saturday, frequent service to and from the conference July 27 at 4:00 PM in the registration area. hotels. The one-way fare is $13.00 CDN; round trip is $20.00 CDN. For more informa- General In General tion, consult www.edmontonairports.com/ Disclaimer gtr/skyshutt.htm, or call 888-438-2342. In offering Avis Rent A Car, Canada Air, Taxi Crowne Plaza Chateau Lacombe, The Fairmont Taxis are available at Edmonton International Hotel Macdonald, Shaw Conference Centre, Airport. The approximate fare from the air- University of Alberta, The Westin Edmonton, port to downtown Edmonton is $41.00 CDN. and all other service providers (hereinafter re- ferred to as “Supplier(s)” for the National Con- Bus ference on Artificial Intelligence and the Inno- Greyhound Bus: For information on fares and vative Applications Conference), AAAI acts on- scheduling, call 1-800-661-8747. The down- ly in the capacity of agent for the Suppliers town Edmonton Greyhound terminal is locat- which are the providers of the service. Because ed at 10324 - 103 Street. AAAI has no control over the personnel, equipment or operations or providers of accommo- City Transit System dations or other services included as part of the The Edmonton Transit System includes Buses, AAAI-02/IAAI-02 program, AAAI assumes no LRT (Light Rail Transit) and DATS (Disabled responsibility for and will not be liable for any Adult Transit Service). They can be contacted at personal delay, inconveniences or other damage 496-1611 or www.gov.edmonton.ab.ca/transit. suffered by conference participants which may The fares are $2.00 CDN. The station closest to arise by reason of (1) any wrongful or negligent the Shaw Conference Center is Central and is acts or omissions on the part of any Supplier or located at 100th Street and Jasper Avenue. its employees, (2) any defect in or failure of any Train vehicle, equipment or instrumentality owned, Via Rail – The Via Rail station is located with- operated or otherwise used by any Supplier, or in a 15-minute drive of downtown Edmonton. (3) any wrongful or negligent acts or omissions The toll free number in Canada is, 1-800-561- on the part of any other party not under the 8630. www.viarail.ca. control, direct or otherwise, of AAAI. 36 37 HALL C Poster Boards 40' AREA BOTBALL TOURNAMENT 30' LOBBY p (Exhibit Area) p (Exhibit D320 120 221 D318 D419 D421 30' 118 D316 219 D417 Poster Boards 50' LOUNGE ROBOT 116 D415 217 EXHIBITION 20' D312 D314 D413 40' SERVICE CORRIDOR 213 215 112 114 208 108 309 209 307 106 65' ROBOT ESCALATORS 305 104 205 BLEACHERS COMPETITION 202 203 102 303 301 200 204 206 201100 207

65' 110'

CENTRE

BUSINESS STORE ROBOT WORK AREA WORK ROBOT 50' TRUCK BAYS LOADING DOCKS Assembly Level Ma

38 Meeting Le Meeting vel vel Map Map