Research On High Precision Map Construction And Location Technology Based On GNSS/INS/LiDAR-SLAM

Autonomous navigation and positioning ability is the premise of landing of autonomous products,among which autonomous positioning and mapping are the most critical steps.GNSS(global navigation satellite system,GNSS)and INS navigation system,INS)is the most widely used navigation and positioning system,but the quality of GNSS signal is poor in urban canyon,tunnel,forest and other scenes.Only relying on INS,the positioning error of GNSS/INS integrated navigation system will accumulate rapidly,which leads to the reliability of the positioning of driverless vehicles challenged.At present,most of the solutions are to integrate laser or visual sensor for auxiliary positioning.According to the dependence of mapping and positioning,simultaneous localization and mapping(SLAM)technology is applied to achieve accurate position and pose output and accurate mapping.LiDAR is one of the commonly used sensors for auxiliary positioning.Because of its high frequency and high precision,it is widely used in the research of relative positioning and high precision mapping.In order to realize the autonomous navigation and positioning ability of autonomous vehicle,this paper studies the 3D map construction and real-time positioning based on LiDAR.(1)Aiming at the algorithm of position and pose estimation of laser SLAM front-end odometer,a method of laser odometer calculation based on feature matching is proposed,and the relevant coordinate system,kinematic model and point cloud registration algorithm are described to lay a theoretical foundation for it.At the same time,original point cloud are segmented,feature extraction,motion compensation and feature association to achieve high-precision pose estimation of autonomous vehicle.(2)Aiming at the problem of error accumulation of laser odometer in large-scale scene mapping,combined with the high-precision global positioning performance of integrated navigation system,a high-precision map building algorithm based on GNSS/INS/LiDAR-SLAM is proposed.The front-end laser odometer adopts the method of laser mileage calculation based on feature matching,and the back-end uses GNSS as a global prior information constraint to join the pose graph and combine the closed-loop detection for global optimization,so as to achieve high-precision pose estimation results,improve the accuracy of 3D environment map,and complete the construction of global point cloud map with the initial pose provided by GNSS / INS integrated navigation system.(3)In view of the poor positioning effect of the integrated navigation system in weak GNSS areas such as urban canyon and tunnel,an autonomous real-time positioning algorithm based on global map is proposed.The core idea is to build a high-precision global point cloud map based on GNSS/INS/LiDAR-SLAM,and use the NDT matching positioning algorithm to realize the autonomous positioning.At the same time,the strategy of organizing point cloud map and dynamic loading map by octree is adopted to meet the real-time performance of the algorithm.On the basis of the above research,through the verification of simulation and measured data,this paper compares and analyzes the three-dimensional environment map building algorithm and the matching and positioning algorithm based on the global map,in order to verify the effectiveness and reliability of the algorithm proposed in this paper.