This objective of this research is to develop a computational framework for behavior modeling adapted to the problem of creating complex scenarios in large-scale virtual environments. The project will develop a modeling system based on hierarchical, concurrent state machines that supports the control of reactive behaviors and provides mechanisms for coordinating behaviors to create critical situations and events. A second research thrust is to create authoring tools to facilitate the design of virtual environments and the definition of scenarios in which computer generated agents play central roles. This technology has application in interactive, real-time simulation environments such as vehicle simulation for driver training, virtual prototyping, and experimental studies of driving safety that require dynamic situations and circumstances to be reproduced across trials without overly restricting subject's actions and while maintaining diversity, responsiveness, and spontaneity in agent behaviors.
