An Approach to Graph-Based Grid Map Segmentation for Robot Global Localization

摘要

Pose estimation is an essential task for robot localization and navigation. It is difficult for mobile robot to estimate the correct pose and recover from the incorrect global localization without prior pose information. It is known as the “kidnapped robot problem” in the field of robotics. The map segmentation is a useful preceding task for robot global localization, and it can facilitate the process of scan-to-map matching. This paper presents a novel method of grid map segmentation for mobile robot global localization. By the clustering algorithm, the grid map constructed by the laser scans can be divided into multiple clusters. The clusters of the map are represented by a graph, and then, are merged into different regions according to the connectivity of the graph. The experimental results from the proposed method are more consistent to the truth segmentation as compared to other methods.