| With the continuous development of technology and society,unmanned vehicles are widely used in transportation,industry,military and other fields.Mapping and navigation are the core functions of unmanned vehicles,and laser simultaneous localization and mapping(SLAM)technology based on multi-sensor fusion is one of the important research directions in the field of unmanned vehicles.This paper studies the SLAM and path planning algorithm of laser inertial navigation fusion,and constructs a complete set of 3D mapping and navigation solutions for unmanned vehicles.The main research contents are as follows:First,aiming at the problems of trajectory drift,low positioning accuracy and robustness of traditional laser SLAM,a close-coupled SLAM scheme of laser inertial navigation based on graph optimization is improved.The front-end uses high-frequency inertial measurement unit(IMU)data to correct and compensate the distorted point cloud,uses the geometric angle relationship between the lidar and the carrier to extract the ground point cloud,and uses the combination of European space and angle threshold to cluster and segment the non-ground point cloud.The back end uses the method of graph optimization,and integrates the laser odometer,IMU and loopback detection information based on the sliding window model to complete the construction of 3D point cloud map.Based on the Kitti data set,the traditional method and this method are compared and analyzed,and the experiment shows that this method has better robustness and mapping accuracy.Secondly,aiming at the low efficiency and unsmooth path of D* algorithm in dynamic path planning,the obstacle distance influence coefficient ρ and four-point gradient descent method are introduced.Set the pre-judgment distance for the obstacle,judge the impact of the obstacle on the path planning according to the size of ρ,adjust the path in time,and improve the obstacle avoidance efficiency.Then,the displacement vector of the second path point in the overall path is used to constrain the smoothness of the four-point path,reduce the road turns,and then iterate to obtain the optimal path.At the same time,the improved D * and time elastic band(TEB)algorithms are fused into autonomous navigation modules.In the robot operating system,the simulation experiments of this algorithm and the traditional fusion algorithm are compared.The results show that the improved fusion algorithm has better navigation and obstacle avoidance performance.Finally,the unmanned vehicle experimental platform is built,and the laser SLAM scheme in this paper is verified by selecting different realistic scenes.Compared with the existing methods,the laser SLAM scheme has better inhibition effect on the motion track drift and point cloud ghost phenomenon.At the same time,the Octomap function package has been improved by adding ground filtering to generate grid maps with 3D information.Then,based on this map,the navigation performance of the unmanned vehicle is tested to verify the feasibility and actual performance of the integrated improved navigation function package.The results show that the unmanned vehicle system can efficiently complete the mapping and navigation tasks. |
