Cooperative Mapping using Small FoV LiDARs from Multiple Vehicles

Point cloud maps play an important role in many autonomous driving tasks. Generally, the professional mapping vehicles with various devices are used for mapping. It is inefficient and high cost in the case of large-scale mapping. Meanwhile, with the development of intelligent vehicles, more and more vehicles are equipped with low-cost and small field of view (FoV) Light Detection and Ranging (LiDAR) sensors. If these vehicles could be used for mapping, there is a great possibility to improve the scale and efficiency of mapping. In this paper, we propose a cooperative mapping method based on multiple vehicles each equipped with a small FoV LiDAR, a low-cost Global Positioning System (GPS), and an Inertial Measurement Unit (IMU) to perform an efficient and consistent point cloud map. The proposed solution adopts the multi-sensor fusion method to build the local maps and uses ground alignment and FoV maps to solve the place recognition across different local maps gotten by different vehicles. The comparative experiment in real practice demonstrates the potential and advantages of the proposed cooperative mapping method in terms of efficiency and consistency of the mapping.