A New Hybrid Metric Map Representation by Using Voronoi Diagram and its Application to SLAM

In this paper we present a hybrid metric map representation, which partitions the global map space into a series of local subregions by operating Voronoi diagram on global feature map and builds a local dense map in each subregion. Compared with the existing hybrid metric map representation [1] [2] (HYMM), this new hybrid metric map representation has the following advantages: First, the Voronoi diagram ensures that the map partition result is unique. Second, the local environment contour in each subregion is proven to be described completely by its corresponding local map. Third, benefited from the presented hybrid metric map representation, a shape matching data association (DA) method is presented to deal with the ambiguity problem of traditional DA approaches. An experiment carried out in a corridor environment validates the effectiveness of the new hybrid metric map representation and the shape matching DA method.