The Multiagent Systems Research Group, under the direction of Assistant Professor Tuomas Sandholm, is concerned with designing, analyzing and implementing sophisticated AI systems consisting of multiple agents. Special focus is on the coordination of self-interested agents in open systems. In this endeavor, we use techniques from game theory and other fields of microeconomics, and we extend those theories with normative models of bounded rationality, such as limited computation and communication. This work has also led us to develop new techniques for single agent resource-bounded reasoning, constraint satisfaction, and machine learning.
Applications with self-interested, bounded rational agents are ubiquitous, including vehicle routing among independent dispatch centers; manufacturing planning and scheduling among multiple companies in subcontracting networks; meeting scheduling; scheduling of patient treatments across hospitals; classroom scheduling; planning and scheduling of multi-contractor software projects; digital libraries; multiagent information gathering on the Web; routing and bandwidth allocation in multi-provider multi-consumer computer networks; electronic commerce; and electronic trading of equities, to name just a few.
The MAS Group is part of the Department of Computer Science at Washington University, St. Louis.