A perception and selective attention system for synthetic creatures

Abstract

This paper describes the selective attention and perception system designed for the synthetic creatures of the ANIMUS[1] project. This system allows agents to evaluate objects, other characters, properties, and events happening in their virtual environment. Perceived data is prioritized to direct the attention of the character toward salient stimuli, providing means for action-reaction and conditioned learning, automatic triggering of reflexive responses to certain stimuli, and further optimization of the cognitive processes by creating a more natural and realistic interface. The system can be customized to fit the artistic and conceptual constraints of each character, facilitating the creation of synthetic creatures with complex behavior and personality. © Springer-Verlag Berlin Heidelberg 2003.