Mobile Robot Localization Based on Random Closed Set Model Maps

In this paper we present a novel approach to mobile robot navigation in natural unstructured environments. Natural scenes can be considered as random fields where a large number of individual objects of random shape appear randomly scattered in space. In this paper we use Random Closed Sets (RCS) to model the random scattering and shape of the objects observed, and base the navigation of a robot on maps of the RCS model’s parameters. Contrary to the feature based approach to robot navigation, this environment representation does not require the existence of outstanding objects in the workspace, and is robust with respect to small dynamic changes. We address the problem of estimating the position of a mobile robot assuming that the (statistical) map of the environment is available a priori. We also present an adaptive guidance strategy that autonomously leads the robot to locations where the perceptual observations result in an efficient reduction of its state uncertainty. Simulations demonstrate the feasibility of our approach.