Real-Time Robot Localization Based on 2D Lidar Scan-to-Submap Matching

In this paper, we propose a real-time and low-drift localization method for lidar-equipped robot in indoor environments. State-of-the-art lidar localization research mostly uses a scan-to-scan method, which produces high drifts during the localization of the robot. It is not suitable for robots to operate indoors (such as factory environment) for a long term. Besides, the mapping and localization of this method are susceptible to the dynamic objects (such as pedestrians). To solve above problems, we propose the scan-to-submap matching method for real-time localization. Currently, this method has been used for building maps, and there are few studies to use it for localization, especially for real-time localization. In our research, we build the hardware and software platform for the scan-to-submap matching method. We extensively evaluate our approach with simulations and real-world tests. Compared with the scan-to-scan method, the results demonstrate that our approach can cope with the mapping and localization problem with high localization accuracy and low drift.