Event Location: 1305 Newell Simon Hall
Bio: Dr. Pedro Szekely is a Project Leader in USC/ISI and a Research Assistant Professor in USC’s Computer Science Department. He received his M.S. and Ph.D. degrees in Computer Science from Carnegie Mellon University in 1982 and 1987. Pedro has worked on user interfaces, multi-agent systems, and planning and scheduling. His recent work is focused on user interfaces and algorithms to help human planners construct good plans for complex dynamic and uncertain domains. This talk summarizes the lessons learned in the DARPA COORDINATORS program and the issues involved in taking algorithms that work well in simulation and using them to solve problems in the field.

Abstract: Coordinating multiple agents in the real world is difficult because it requires simultaneously addressing planning, scheduling, uncertainty and distribution. Generic AI approaches often produce inadequate solutions because they cannot leverage the structure of domains and the intuition that end-users have for solving particular problem instances. We present an approach where end-users encode their intuition as guidance enabling us to decompose large distributed problems into simpler problems that can be solved by traditional centralized AI techniques. Independent evaluations in real world field exercises with human users show that teams assisted by our multi-agent decision-support system outperform teams coordinating using radios.