Performance of decision theory rules in a multiagent environment

This research addresses the problem of searching for objects of interest in a dynamic environment with a team of autonomous Agents. We will first demonstrate that Bayesian reasoning can be used to both create an accurate map of object locations and guide the Agents' behavior to most effectively maintain the map, even in the face of a dynamic environment and less-than-perfect information. We will then develop performance criteria to gauge the accuracy of the map and the coordination of the Agents that verify that these decision theoretic methods work. Finally, we will provide a web-launchable application program that allows a user to observe the behavior of the Agents in a simulated environment. The simulation program is object-oriented, event-driven, and scalable to any number of Agents and any number of different object types.