Intelligent Virtual Agents Architecture in Unknown Environment

As the contemporary performing behavior in the reciprocity of virtual agents is glazed excessively and can't satisfy complex unknown environment, the research of Intelligent Virtual Agents (IVA) with active learning and local apperception must be accelerated. In this paper, we design and implement a new IVA system architecture, for which active learning and local apperceive interaction were approached. Active learning using knowledge acquisition and knowledge evolution improved the self-regulation of behavior mechanism significantly, while local apperceive interaction could sense disturbances of the simulated environment and interactively communicate with other simulated agents within perception dimension effectively. Ultimately, the experimental result provides effectiveness and practicability of the performance in our prototype IVA system.