In Purves, R. & Weibel, R., editors, Proceedings of the 6th International Conference on Geographic Information Science (GIScience 2010), 2010. Paper abstract bibtex
We investigate environmental learning and navigation capabilities based on qualitative observations and formal spatial reasoning. Our approach is based on previous work in the area of qualitative navigation (Davis 1983, Schlieder 1993, Wagner et al. 2004) in which a classification of perceptions into qualitative categories is used to derive a decomposition of space leading to a graph-based representation of the environment called topological map. As described in Kuipers (2000), the use of topological map is motivated by a cognitively inspired approach to robot navigation that requires to execute a sequence of local behaviors in order to reach a target position. In Fogliaroni et al. (2009) we described a novel qualitative approach able to explore, navigate and build a topological map of an unknown environment based on a model of visibility between extended objects. We now extend the visibility model into a qualitative spatial calculus to apply qualitative spatial reasoning techniques (Cohn and Hazarika 2001). This allows us to employ formal spatial reasoning to improve map learning and navigation.