Multi-Hypothesis Topological Mapping Using Qualitative Spatial Reasoning

We describe a multi-hypothesis mapping system for mobile robots that learns graph-based topological representations. Our approach exploits direction information and the assumption of planarity to prune the space of possible map hypothe- ses. Qualitative spatial reasoning is used to check satisfiability of individual hypotheses. We evaluate the effects of ab- solute and relative direction information and incorporate the approach into a mapping system based on Voronoi graphs.