The Data Association Problem

Abstract

“It is said that one image is worth a thousand words.” In order to reach a consensus, the first required step for the team of robots is to globally identify these “words.” In this chapter we address the problem of finding global correspondences between the observations of all the robots in a distributed manner. At the beginning, each robot finds correspondences only with the robots that can directly communicate with it. This is done using existing matching techniques for pairs of images. After that, we study a distributed algorithm that propagates the local correspondences through the network. We formally demonstrate the main properties of the algorithm and prove that after executing our method, the team of robots finishes with a globally consistent data association. The performance of the algorithm is tested with extensive simulations and real images at the end of the chapter.