Programmable Agent Perception in Intelligent Virtual Environments

In this paper, we propose a perception model for virtual agents based on a two-layered architecture. Each agent is equipped with a sensing mechanism that retrieves the objects in its field of view and stores them in its memory, and with a perception module that applies a set of rules on them and generates its beliefs. We propose a scripting language for defining such perception rules using condition->assertion statements, and describe in detail an example of a perception case.