Reasoning about Action for Behavioral Animation of Intelligent Virtual Agents

—Recent work in behavioral animation has taken impressive steps towards a virtual human. As for intelligent virtual agents, abbreviated " IVAs " , the behaviors displayed in virtual environments should be " reasoned ". It remains challenging, however, to endow virtual agents with cognitive ability to perform specific tasks. This paper addresses cognitive models and behavioral models for IVAs. For cognitive models, concurrent actions, complex actions, control structure, and corresponding semantics are added to FLUX software that is based on fluent calculus to create ConPFLUX software. It enables IVAs to reason about action in terms of given tasks in dynamic and incompletely known environments. For behavioral models, we develop a ConPFLUX-based animation description language FCADL. It is capable of conveniently and flexibly describing the animation of various complex behaviors of IVAs resulting from ConPFLUX.