Study On The Inertial Navigation System Aided By LiDAR Plane Registration

With the development of mapping technology,the mapping method is changed from fixed station to mobile device,and the mapping scope is also extended from outdoor to indoor.Mapping technology is inseparable from positioning technology.Compared with the mature positioning solutions relying on navigation satellites outdoors,indoor positioning environment is more complex and the positioning is more difficult to achieve.In this paper,with the laser radar scanning rangefinder and inertial measurement unit as the main sensor,the self-developed mobile mapping backpack is used as the platform to explore the positioning and navigation scheme in the unknown structured indoor environment,respectively from the theory and practice.First of all,the paper analyzes the error model of inertial devices used in project and quantify the relative error term.Based on the error model of inertial device,the process of inertial navigation algorithm and the update equation of the system error state are designed.Then the coarse alignment and the precision alignment calibration method for the attitude initialization are introduced.Finally,through the footsteps track experiment,the validity of the algorithm of inertial navigation is proved.Then,the accuracy and distribution characteristics of point cloud acquired by laser radar scanning rangefinder used in the subject are analyzed.And based on this,the point cloud plane registration algorithm is designed,including the point cloud posture prediction,utilizing region growing method and random sampling method to extract the point cloud surfaces,the calculation of point cloud plane parameters and the corresponding covariance,the choice of the corresponding planes and the calculation of relative position and rotation between laser scans.In the plane extraction experiments,the effect and efficiency of plane extraction algorithm used in this subject are compared with other algorithms.In the plane registration experiment,the advantages and disadvantages of the plane registration algorithm and the traditional point cloud registration algorithm are compared.Finally,with the time synchronization and space calibration of the inertial device and the Li DAR,the displacement and velocity of the Li DAR obtained by point cloud registration is used as the observation value of extended kalman filter,and the indoor combined navigation system is designed.The accuracy and robustness of the system are evaluated through experiments.The results show that the proposed algorithm can effectively limit the error drift of the inertial navigation system and achieve stable and reliable indoor positioning and navigation.