Invited Speakers

HUGH DURRANT-WHYTE

Title: Control in Decentralised Sensor Networks (Joint work with Alex Makarenko, Tomo Furukawa, Ben Grocholsky)

• Abstract: Information gathering tasks such as exploration, navigation and searching provide a particularly interesting case of multi-robot control and coordination. Crucially, such tasks are characterised by having integral pay-off, in which the set of robots receive a continual measure of value, and by being decoupled from physical interactions through the environment, as sensing actions do not generally affect the state of the environment. These characteristics make solutions to the multiple sensor information gathering problem far more tractable than the more general multi-agent control problem.  This presenta tion will describe an information-theoretic approach to problems of control in decentralised sensing networks. The approach builds on much past work in decentralised data fusion by using ideas of information gain as the  key metric for controlling and coordinating the actions of different sensors. Operations such as exploration, navigation and search can be described in terms of maximising information gain, employed as both local and global objectives for sensing agents. Coordination between agents is facilitated because each agent shares a common sensed model of the environment. Three control regimes for information gathering are described, only one of which requires explicit coordination of sensing actions. A greedy algorithm for information maximisation which requires minimum communication is demonstrated in problems of tracking and exploration. A key feature of this method is that it scales well to large numbers of agents dynamically entering and leaving the global sensing task. Demonstrations on indoor robot vehicles equipped with laser and vision sensors is described. Simulation results and planned experimental work on a fleet of unmanned air vehicles (UAVs) will also be described. In conclusion we describe the extension of some of these ideas to terminal pay-off functions where the problem of value assignment to a group is of significant concern. These include problems of rendezvous and engagement.

• Brief Biography: Hugh Durrant-Whyte received the B.Sc. (Eng.) degree (1st class honours) in Nuclear Engineering from the University of London, U.K., in 1983, the M.S.E. and Ph.D. degrees, both in Systems Engineering, from the University of Pennsylvania, U.S.A., in 1985 and 1986, respectively. From 1987 to 1995, he was a Lecturer in Engineering Science at The University of Oxford U, From 1995 to 2002 he was the Professor of Mechatronic Engineering in the School of Aerospace, Mechanical and Mechatronic Engineering, the University of Sydney, Australia, where he led the Australian Centre for Field Robotics (ACFR), a Commonwealth Key Centre of Teaching and Research. In 2002 he was appointed an ARC Federation Fellow and is now also the Research Director of the ARC Centre of Excellence in Autonomous Systems at The University of Sydney. His research work focuses on autonomous vehicle navigation and decentralised data fusion methods. His work in applications includes automation in cargo handling, surface and underground mining, defence, unmanned flight vehicles and autonomous sub-sea vehicles.

NICHOLAS R. JENNINGS

Title:Negotiation Technologies

• Abstract: Negotiation is a key form of interaction in a wide variety of areas (including multi-agent systems, the Grid, pervasive computing, and the Semantic Web). Given this ubiquity, automated negotiation technologies exist in many different forms, each of which has different characteristics and properties. Against this background, this talk discusses work on a variety of models, covering bi-lateral encounters, auctions and argumentation-based negotiation.

• Brief Biography: Nick Jennings is Professor of Computer Science in the 5*-rated Department of Electronics and Computer Science at Southampton University where he carries out basic and applied research in agent-based computing. He is also the Chief Scientific Officer for Lost Wax. He helped pioneer the use of agent-based techniques for real-world applications, has been a leading figure in the field of agent-oriented software engineering and has made foundational contributions in the areas of automated negotiation and auctions, cooperative problem solving, and socially rational decision making. He has published some 180 articles and 6 books on various facets of agent-based computing and holds 2 patents (3 more pending). He is in the top 125 most cited computer scientists (out of 660,000) according to the CiteSeer digital library and has received a number of awards for his research: the Computers and Thought Award (the premier award for a young AI scientist) in 1999 (this is the first time in the Award's 30 year history that it has been given to someone based in Europe), an IEE Achievement Medal in 2000, and the ACM Autonomous Agents Research Award in 2003.

W. LEWIS JOHNSON

Title: Social interaction with agents

• Abstract: Social psychologists such as Reeves and Nass have argued that people tend to relate to media much as they relate to human beings. Animated synthetic agents have been developed to exploit this tendency. Such agents raise expectations that they can function as social actors able to engage in social interactions with people and other agents. This presentations will describe efforts at enabling synthetic agents, particularly pedagogical agents, to interact in a manner that is sensitive to social expectations and conventions. The social intelligence that this requires is of practical value in promoting learner motivation and engagement. Finally I will discuss efforts to enable agents to act in a dramatic sense to portray roles and convey emotion. This is an essential part of human social interaction, which often involves the portrayal of roles. And the application of theatrical principles to agent behavior results in agent behavior that is more consistent and understandable, and user experiences that are more engaging.

• Brief Biography: W. Lewis Johnson is director of the Center for Advanced Research in Technology for Education (CARTE) at the University of Southern California's Information Sciences Institute. His work involves the application of artificial intelligence to education, training, and human-computer interaction generally. He is past chair of the ACM Special Interest Group for Artificial Intelligence, and past president of the International Artificial Intelligence in Education Society. He served as co-chair of the First International Conference on Autonomous Agents, program co-chair of the First International Joint Conference on Autonomous Agents and Multi-Agent Systems, and program co-chair of the International Conference on Intelligent User Interfaces.

DAVID PARKES

Title: Computational Mechanism Design: Taming the Strategic Dragon without Invoking the Complexity Monster

• Abstract: Computational systems are increasingly open, and designed, operated, and used by multiple autonomous and self-interested parties. Witness the recent explosion in peer-to-peer, e-commerce, grid-computing, and on-demand computing. Mechanism design (MD) provides a beautiful mathematical theory with which to design rules for these multiagent systems and provide incentives to promote good behavior despite the individual self-interest of agents. Yet, MD is embarrassingly brittle and impractical in its rawest form, invoking centralized complexity and complete revelation of private information to tame the dragon of game-theoretic complexity. Against this backdrop, this talk discusses computational issues in the deployment of mechanisms in MAS, covering the imposition of approximations, the role of agent bounded- rationality, and dwelling in particular on an agenda in distributed mechanism design.

• Brief Biography: David Parkes is an Assistant Professor in Computer Science at Harvard University where he carries out research on topics at the interface between computer science and economics, and in particular in areas of mechanism design, e-commerce, multiagent systems, and game theory. He received his Ph.D. degree in Computer and Information Science from the University of Pennsylvania in 2001, and an M.Eng. (First class) in Engineering and Computing Science from Oxford University in 1995. A recent recipient of the NSF CAREER Award for his work in computational mechanism design, Dr. Parkes had earlier received the IBM Institute of Advanced Commerce award for best dissertation proposal in electronic commerce. He serves on the Program Committee of a number of leading conferences in artificial intelligence, multiagent systems, and electronic commerce, and has helped to organize a number of smaller symposia. Dr. Parkes is the author of a number of patents, and serves as a technical advisor to CombineNet Inc., a start-up company that specializes in advanced optimization software for electronic markets.