Data integration in multi-sensor based robotic workstations

Abstract

A relaxation labeling algorithm is developed. The major advantage of this algorithm over the existing ones is that the mathematic operation is simplified. The simplification eases the analysis of the convergence properties. Both the theoretical and application aspects of the proposed algorithm are investigated. The local convergence properties of a labeling process with n labels and m labels are established. The investigation of the interaction among the nodes in a multinode labeling process reveals some insight into the mathematical issues involved in the relaxation operations.