Topological localization on indoor sonar based fuzzy maps

In this paper, a topological graph is used as a world representation for a mobile robot that navigates in an office-like environment. The nodes of the graph represent the intrinsic structure of the environment like corridors, corners and so on. The arcs capture the connectivity of the space. We propose a new algorithm to identify the characteristic features of nodes directly from sensory data, during the robot navigation. This is achieved by snatching the idealized and the real patterns of the feature and by taking into account the previous results through the Transferable Belief Model. This information can be used to localize the robot on a given graph or to build a topological representation of the environment.