Modern autonomous systems are challenged by a necessity of the permanent situation awareness: perception, comprehension, and prediction of the surrounding environment. A basis for such awareness is a reliable world modelling that serves as an efficiently structured memory.
This contribution proposes several new approaches in the world modelling. The Progressive Mapping represents a dynamical description of real world elements with mapping sets of objects and attributes. The prior knowledge about object types and relations is introduced as a collection of semantic networks of classes, while dynamic relations within the world model are represented by semantic networks of objects. Also, current contribution presents a processing of degree-of-belief distributions and a possibility of calculation of memory limits.