Communicating with Animated Agents in Behavioral Languages

摘要

This paper presents a novel approach to computer animation of articulated characters that can be interactively controlled and trained by users through a high-level behavior-oriented language. The graphical agents are capable of associating a behavioral pattern with a particular behavioral command, and most importantly given the specific local conditions of the virtual environment, adapting their exact configurations of the activated behavioral patterns, through an evolution-strategy based optimization. Furthermore, the conditioned behaviors will be recorded which can in turn serve as a guidance for future behavioral evolution. Inrnthis paper, we give an overview of our system, describing how users can graphically create an animated agent and behaviorally command it. We then describe the kernal of our animated graphical agent system which dictates the adaptation and learning in the animated agents. Finally, we present an illustrative example of the behaviorally operated and adapted graphical agent.