Indoor navigation by using segmentation of range images obtained by laser scanners

Abstract

This paper proposes a method to reconstruct the trajectory of a moving laser scanner on the basis of indoor scans made at different positions. The basic idea is that a position with respect to a measured scene is completely determined by the distances to a minimum of three intersecting planes within that scene, and therefore displacements are defined by changes in those distances between successive scans. A segmentation algorithm based on the gradients of a range image is used to extract planar features from the measured scene. The correspondence establishment between planes of two successive scans is formulated as a combinatorial problem, and a heuristic search method is employed to find co..