An Environment-Augmented Global Localization Solution Using LiDAR and Geo-Referenced Point Cloud

Accurate global localization is still a huge challenge in complex condition, where satellite signals are obscured. Fusing LiDAR and geo-referenced point cloud is a feasible way to obtain accurate global locations in global navigation satellite system denied environments. However, the issues of limited generalizability and low accuracy remain with existing research using geo-referenced point cloud from various data collection systems. To solve the above problems, this paper utilizes virtual LiDAR in the preprocessing procedure to generate general global prior map using the existing geo-referenced map, overcoming the generalizability limitation of the existing methods. At the same time, the proposed solution adopts an environment-augmented and LiDAR odometry checking and tracking strategy to improve global localization accuracy. Comprehensive experiments confirm that the proposed method shows sufficiently improved performance.