Reserch And Implementation Of Unmanned Driving Relocation Based On Multi-sensor Fusion Of LiDAR And GPS

Multi-sensor fusion can solve the nonlinearity and uncertainty of sensors.Aiming at the problem that GPS and SLAM cannot provide accurate positioning for unmanned vehicles in complex environments,a relocation algorithm based on point cloud map matching is proposed.Fusion of map matching positioning and GPS positioning results to obtain accurate pose information of unmanned vehicles.This article is mainly divided into 4 parts:(1)Based on the unmanned driving platform,build a laser radar-based relocation hardware and software system.Three-dimensional lidar data processing and multi-sensor synchronization algorithms are studied,including laser sensor position calibration,point cloud data noise filtering,point cloud distortion correction,and multi-sensor data synchronization.Finally,multiple sensors are realized in time and space dimensions.Synchronize.(2)Propose an efficient map construction and optimization algorithm for the creation of high-precision point cloud environment maps of closed parks.Use the improved adaptive ICP algorithm to construct the front-end odometer,and use the algorithm based on pose graph optimization for back-end optimization.Use the laser odometer to construct the pose map and add ground constraints,GPS prior pose constraints and detected closed-loop constraints,optimize the pose sequence according to the constraints,use the optimized pose sequence to stitch key frames,and complete the increment of the global point cloud map Style construction.(3)Propose an unmanned vehicle relocation algorithm based on a priori map to obtain the absolute position of the unmanned vehicle on the map.GPS provides the initial position,uses map matching to update the pose observation,GPS data for pose prediction,and uses the unscented Kalman filter to fuse the positioning results based on map matching with the GPS positioning results,which improves the positioning accuracy.(4)Based on the unmanned driving platform,collect real data from campus semi-structured scenes to verify the feasibility of the algorithm.Experiments show that the accuracy and real-time performance of the positioning algorithm based on the fusion of lidar and GPS multi-sensor can meet the positioning system of unmanned vehicles.demand.