Simulating UAV Surveillance for Analyzing Impact of Commitments in Multi-Agent Systems

Abstract

Autonomous agents, by definition, have the freedom to make their own decisions. Rational agents execute actions that are in their "best interests" according to their desires. Action selection is complicated due to uncertainty when operating in a dynamic environment or where other agents can also influence the environment. This paper presents an action selection framework and algorithms that are rational with respect to multiple desires and responsive to changing desires. Coordination is layered on top of this framework by describing and analyzing how commitments affect the agents' desires in their action selection models. Commitments may have a positive or a negative effect on an agent's ability to satisfy its desires. This research uses simulation in the domain of UAV surveillance to experimentally explore the balance between under-commitment and over-commitment.