Exploration and Navigation Using Hierarchical Cognitive Maps

We describe a spatial navigation model that uses a hierarchical representation of space. The model is able to explore exhaustively large environments and to plan paths between distant places, even though it uses a working memory with a limited capacity. By using a hierarchical structure, an agent with a limited working memory capacity has (a) a comparable performance to that of an agent with no working memory limitations, (b) a reduced number of connection values and (c) a shorter decision time. The model describes many properties of hierarchical spatial behavior in humans and animals. Potentially, the approach can also be applied to the design of robots able to navigate in large environments. Chapter Outline & Navigation