Mobile Robot Localization Using Map Based on Cadastral Data for Autonomous Navigation

For autonomous navigation, localization in an environment is a fundamental capability. In this regard, it is necessary for robots to have environmental maps. The environmental maps are, in general, built beforehand through SLAM. On the other hand, the authors have focused on cadastral data including road and building information. In this paper, we propose a localization method using an environmental map based on the cadastral data. The robot is assumed to be equipped with 3D LiDAR. However, the cadastral data does not include natural objects, such as trees. Moreover, the position of a building in the cadastral data is sometimes different from its actual position in the environment. For data association between the sensor measurements and environmental map, we describe point cloud processing. Furthermore, we use an RGB-D camera to obtain a dense point cloud. In the navigation experiments, we show that the robot is able to move towards the destination autonomously through the localization in the map.