Voronoi Graph Matching for Data Association

Abstract

In this chapter, we will be concerned with the problem of identifying correct correspondences between an AGVG extracted from local information about the robot’s surroundings and the memorized map AGVG. Identifying the correct correspondences is a crucial step for complementing the map AGVG incrementally by inserting newly perceived parts from the local AGVGs and for localization in general. We develop a matching approach based on the notion of edit distance that compares the graph structures and the annotations of the AGVGs. Results from the previous chapter with regard to assessing the stability of Voronoi nodes are incorporated into the matching process in order to adequately deal with potential variations in the AGVGs.