Research On GNSS/INS/LiDAR High Precision Positioning Method In Weak GNSS Signal Regionand Its Application

GNSS navigation and positioning technology is the main means of outdoor high precision positioning,the GNSS signal will not be blocked and the quality of observation is good in the open environment.At this time,we can achieve dynamic centimeter level positioning accuracy by the use of RTK(Real-time Kinematic).Mobile mapping system projected the information gathered by environmental remote sensing sensors(LIDARs,cameras,etc.)into a unified framework to achieve high-precision mapping by the use of GNSS/INS position sensor and attitude sensor,and it has developed rapidly while GNSS high-precision positioning technology is widely used in various fields.However,in the forest,urban,canyons and other regions in which the GNSS satellite signals often are blocked by shelter or building and the observation error increases as well as the signal strength be weaken.The positioning accuracy will degenerate from centimeter level to decimeter level even meter level because the ambiguity resolution is poor in RTK and the positioning accuracy which can't meet the needs of high precision mapping also can't be effectively improved even through the GNSS/INS integrated navigation system.In order to solve the problem of improving the positioning accuracy in the weak GNSS signal region,this paper introduces-the Lidar feature matching with high positioning accuracy to combine with GNSS/INS and realizes the high precision positioning in the weak GNSS signal region.This paper analyzes the GNSS/INS navigation theory,LiDAR navigation theory,GNSS/INS/LIDAR multi-sensor fusion theory and so on to research the feasibility of high-precision positioning in the weak GNSS signal region and to provide a new solution for the continuous stable and high-precision positioning of applications such as mobile mapping.Aiming at the existing problems and deficiencies in related research,a new combined navigation model of GNSS/INS/L-IDAR is proposed,and a practical solution of centimeter-level dynamic positioning for long distance is given as well as the feasibility and effectiveness of the solution was verify through practical application.The main work of this paper is as follows:(1)Draw a conclusion about the basic theory of integrated navigation,including common coordinate system and the expression of its transformation relation,the observation model of sensor,the measurement model of GNSS and LiDAR,space-time synchronization method,Kalman filter theory and its extended theory.A high-precision INS mechanization is derived and achieved under navigation coordinate system and the GNSS/INS loose coupling extended Kalman filter model is constructed.Finally,the navigation performance of GNSS/INS integrated navigation in different environment is analyzed through simulation.(2)The LiDAR SLAM theory based on probability was summarized and the EKF-SLAM model based on point feature was deduced and established in this paper.Furthermore,we analyzed the relative positioning accuracy of EKF-SLAM based on the simulation that indicated that the relative positioning accuracy of EKF-SLAM which only can be maintain in short distance decreases with the running distance increases,and the main reason for the error of EKF-SLAM position is heading angle error.Sensitivity analysis provides the method to the relative positioning accuracy of EKF-SLAM from three aspects,including improving the observation accuracy of landmark,improving the effective observation distance of LiDAR and providing heading angle information to EKF-SLAM.The first two methods need to improve the hardware cost and the third method with the relative positioning accuracy is obviously improved can be realized by combining with other navigation sources.It's possible to achieve high-precision relative positioning within long distances if we attached EKF-SLAM with high-precision heading angle constraint information by other observation source.We derived the LiDAR/ESNS combined navigation model based on the navigation coordinate system and the simulation indicated that the relative positioning accuracy of LiDAR/INS combined navigation improved significantly compared with single system by the use of tactical IMU.It's difficult to achieve long-distance high-precision relative positioning rely on the combination of INS and LiDAR alone because of the high demand of the heading angle initial accuracy,and the relative positioning accuracy of combined system will decline quickly if there is an error in the initial heading angle.(3)We build the tight coupling of GNSS/INS/LIDAR integrated navigation model of which the navigation system is ENU based on the dynamic and static filtering model,that method not only overcomes the shortcomings of the low fault tolerant of the centralized Kalman filter,but also improves the disadvantages of the suboptimal estimation of the federated filter,and is easy to combine with the robust filter and the adaptive filtering theory.We analyzed the navigation performance of GNSS/INS/LIDAR navigation which can significantly improve the heading accuracy and position accuracy of GNSS/INS by simulation experiment.The GNSS/INS/LIDAR baserd on tactical level IMU can realize centimeter-level dynamic positioning over long distances in the weak GNSS signal area.According to the characteristics of the weak GNSS signals area,we put forward a distributed EKF-SLAM model that also called distributed EKF-SLAM to realize the combination of the three.Combining the navigation results of GNSS/INS with the characteristic observation of Lidar in the frame of EKF-SLAM and distributed processing GNSS/INS and EKF-SLAM.Not only can we improve the short-term relative positioning accuracy,but also reduce the system complexity,improve the stability,and easy to achieve in engineering compared with loosely coupling model.This paper explores the use of high accuracy relative positioning results of combined system to assist GNSS realize the fast re-initialization method under long-term interruption,and constructs the Inter-generational differential model,and adopts the ambiguity change search based on damping LAMBDA.Experiments show that this method can effectively reinitialize the estimation of ambiguity change under long-term interruption of GNSS and achieve rapid re-initialization(4)The theoretical analysis of this paper is carried out through the forestry mapping.It mainly includes the following three points:GNSS/INS integrated navigation is a traditional position and orientation method of mobile laser measurement system,but we can't get RTK fixed solution in forest in which GNSS signal will be blocked even though using high-precision GNSS receiver and tactical level IMU.The positioning accuracy of GNSS/INS integrated navigation is only sub-meter level,but the attitude accuracy which determined by the IMU still maintaining a high accuracy and post-processing heading angle RMS can reach 0.07 °in the experiment of this paper.EKF-SLAM navigation is an effective means of navigation in dense forests,but the introduction of observation errors due to the absence of environmental characteristics will lead to errors in position and heading angles which will cause position-errors accumulated quickly.EKF-SLAM can reach the cm level dynamic positioning in short distance in the case of rich features when the distributed EKF-SLAM can achieve the cm level relative positioning accuracy throughout the experimental section,because the relative positioning accuracy of EKF-SLAM be well assisted by the heading angle and velocity information of GNSS/INS and the short-distance relative positioning accuracy compared to EKF-SLAM also be significantly improved.This method can achieve cm level dynamic positioning accuracy in long distance under the forest environment for the first time,and the short-term relative positioning accuracy also higher than the GNSS/INS,which effectively improve the quality of mapping.All in all,the long-range and high-precision positioning method in the weak GNSS signal area based on GNSS/INS/LIDAR integrated navigation was researched in this paper,and we propose a new solution which feasibility and effectiveness was verified through forestry mapping application from the theoretical and practical point of view,...