Autonomous mapping using a flexible region map for novelty detection

Abstract

This paper presents an autonomous method for a robot to map the normal condition of its environment using a flexible region map. The map is used as a reference to allow a mobile robot to perform novelty detection. The map has a flexible structure which could accommodate to the distribution of different types of entity in the environment. However, updating information in the map for autonomous mapping is not a trivial task since it requires changing the structure of the map. The contribution of this paper is twofold. First, the method for reshaping a flexible region map is discussed. Then, an algorithm that is inspired by the habituation principal for performing autonomous update is presented. Experimental results show that autonomous update was achieved by using the habituation principal and by allowing the flexible region to reshape itself to accommodate to changes in the environment. © 2009 Springer-Verlag Berlin Heidelberg.