A Graph-Theoretic Approach for Constraining Floor Plan Estimation from Radar Measurements

This paper addresses the problem of floor-plan estimation using a limited number of stand-off radar aspect angles, time-delay measurements and wall-penetrations. While most floor-plan estimation work has focused on local methods for reconstructing wall positions within a building, i.e. imaging methods, this paper approaches the problem by incorporating global constraints into the solution. This is done by associating each room in the building with a node on a directed graph that encodes its adjacency to the other rooms. Floor plan reconstruction methods are presented for both cases where the topology of the building is known or unknown.