When designing virtual humans, it is imperative that the virtual agent behave in a human-like fashion. In certain circumstances, this requires agents to be bounded in their ability to sense and understand the environment. To this end, we create a methodology that provides perceptual attention based on a linear combination of senses. We use different heuristics on the object and events that can be sensed by a virtual agent, and combine these scores to create an overall score for a given object or event. This allows the virtual agent to perceive interesting or unexpected items. To demonstrate our technique, we give an example showing the ability of using a linear combination of senses.
展开▼