Local terrain mapping via 3D laser based bubble surfaces

This paper presents a novel approach to local terrain mapping with topological maps. In particular, we use bubble space representation which has been previously proposed for representing nodes in topological maps. We show that bubble surfaces are constructed using three-dimensional laser data are easily associated with slope maps that define local terrain slope and its proximity for each different pan direction. Thus, instead of constructing a global terrain map a priori, a local terrain map is estimated on-demand in real-time as the robot navigates to a new place and constructs a bubble surface. We present experimental results with simulated and real datasets that serve to validate this approach.