Utility Fusion: Map-Based Planning in a Behavior-Based System

* New address as of November 3, 1997: Dept. of Mechanical & Mechatronic Engineering University of Sydney, NSW 2006, Australia A new means of distributed action selection via utility fusion is introduced here as an alternative to both the bottlenecks of centralized systems and the incoherent action selection of distributed systems. Within the utility fusion framework, distributed, independent, asynchronous behaviors indicate the utility of various possible world states. A centralized arbiter then combines these utilities and determines the next action based on the maximization of expected utility. The centralized fusion of utilities from distributed decision-making processes thus provides the responsiveness and robustness of behavior-based systems without sacrificing the coherence and rationality of centralized architectures.