Localizability based path planning on occupancy grid maps
Takuma Nakahara, Yoshitaka Hara, Sousuke Nakamura · Advanced Robotics · 2024
In this paper, we propose a path planning method with graph search considering localizability. Localizability is a measure of the reliability of localization. This paper aims to plan paths that avoid places with low reliability of localization by considering localizability. In the proposed method, we constructed a graph based on occupancy grid maps and planned optimal paths by graph search. Localizability costs were considered as path costs for graph search. In our experiments, path planning was performed in three environments: a corridor environment with parallel walls, a square environment larger than the sensor measurement range, and a real environment. Experimental results confirmed that the proposed method planned paths to avoid places with low reliability of localization, such as parallel wall corridors or the center of squares where the sensor scans cannot be obtained. In addition, we conducted comparative experiments between the paths planned by our method and those planned without considering localizability costs. We confirmed that the paths of the proposed method have lower localizability costs and lower localization errors.