A perception framework for intelligent characters in serious games

The use of BDI-agents seems a good fit to realize intelligent behavior for virtual humans. One of the problems of the BDI-paradigm when an agent becomes embodied in a virtual environment is the lack of control over perception [3]. While performing a task, humans tend to direct their attention to selected information from the environment which can support them in achieving the task. As attention is considered to be a limited resource, one cannot attend to all aspects in the environment which currently fall into sensory range. The same can be said for BDI-agents. Without any form of goal-directed perception, an agent can become flooded with sensory information from the virtual environment, which may result in reasoning over too much irrelevant information. Besides the risk of performance loss this is also unrealistic when we look at the physiology of human perception [1]. A balance must be found between stimulus-driven and goal-based control over perception. In this paper we present a perception framework which provides sensing abilities and perceptual attention for BDI agents embodied in a virtual environment. The framework handles covert attention, the mental focus on possible sensory stimuli which doesn’t involve any motor actions [2]. Different perception stages are identified together with the information communicated between the stages. We show the advantages of using ontological data representations for this Cite as: A Perception Framework for Intelligent Characters in Serious Games (Extended Abstract), Joost van Oijen and Frank Dignum, Proc. of 10th Int. Conf. on Autonomous Agents and Multiagent Systems (AAMAS 2011), Tumer, Yolum, Sonenberg and Stone (eds.), May, 2–6, 2011, Taipei, Taiwan, pp. 1249-1250. Copyright c © 2011, International Foundation for Autonomous Agents and Multiagent Systems (www.ifaamas.org). All rights reserved. information to form an agreement between a BDI-agent in a multi-agent system and its embodiment in a game engine. To illustrate its use in the perception framework, we present an approach for implementing goal-directed perception for BDI-agents.