This report focuses on extending the object-oriented world model by a behaviour model for its representatives. The world model in general is used as a foundation for fusing multiple sensor sources into one coherent picture. It should enable other services to access the stored information for further processing, e.g. for recognizing suspicious situations in surveillance tasks. While the base model is able to capture the real world entities by translating them into representatives while incorporating background-knowledge in form of concepts, it is not able to predict the behaviour of these representatives. Here, a concept based on intelligent rational agents is introduced.