We propose an approach that propagates distance and orientation intervals to integrate imprecise local knowledge into survey knowledge. In the context of mobile robot exploration this propagation scheme can be used to reason about the positions of landmarks or other salient positions and thus can be utilized to generate hypotheses for cycles in the environment. We describe a path-based and mainly topological representation and navigation approach based on the generalized Voronoi graph and demonstrate how the propagation method can be applied in this context.