This contribution presents a new approach for an object oriented environment model which is designed to provide an autonomous system with permanent knowledge on its environment. The world model also acts as an information hub for sensory and cognitive processes. For inserting, updating and removing information about the world, methods of probability theory are applied to the objects. The model is built following three main rules: all information is associated to object and relationship instances; each information in the model is accompanied by an uncertainty statement in a probability notion interpreted as degree-of-belief (DoB); and the handling of the information along with the mechanism ensuring the model quality are based on Bayesian fusion methods.