Efficient Local Navigation Using 1 D Image Representations

In this paper a new, e�cient approach to accurate image-basedhoming is proposed for indoor robots that makes use of the ceiling plane.A gradient orientation histogram is built to calculate the rotation ofthe robot between a target location `snapshot' and the current imageperception, and e�cient 1D average row and column images are constructedto nd the vector pointing to the home location. Our algorithmis ideally suited for use on low-powered, cheap robots where sensing andcomputational power are limited.