Research On Autonomous Navigation Method Of UGV Based On Inertial/Lidar

With the continuous development of robotics,mobile robots gradually appear in our daily lives and play a very important role.As a key technology in the intelligent development of mobile robots,navigation can provide robots with accurate position information when they perform complex tasks.Autonomous navigation based on inertial/lidar is a technology that uses IMU and lidar sensors as navigation information sources to achieve autonomous positioning in an unknown environment.It does not rely on external information interaction and can operate stably in an environment without light.Suitable as a navigation system for mobile robots.Therefore,this article focuses on the inertial/lidar autonomous navigation system for unmanned vehicles from the following aspects.First,the inertial sensors and lidar sensors involved in the autonomous navigation technology are introduced separately,the coordinate system used in the autonomous navigation system is defined,and the relative conversion relationship between the coordinate systems is given.The measurement models of two kinds of sensors are given,and how to use a single sensor to realize its own positioning is introduced.Using the pose information calculated by IMU in real time,the motion distortion existing in the original point cloud of the lidar is removed,and the de-distorted point cloud data is preprocessed,and features are separated from the point cloud data by calculating the curvature.On this basis,and then achieve the matching of feature point pairs in the two frames of point cloud data.On this basis,a lidar odometry is constructed.Aiming at the problem of poor accuracy of lidar odometry,the algorithm based on feature point matching is improved on this basis,and a factor graph optimization algorithm is added to the backend optimization,which can realize the tight coupling of IMU and lidar data;The relatively large difference in the relative movement distance between the lidar factor key frames leads to the problem of poor lidar-frame matching accuracy.An algorithm to remove the non-coincident point cloud of adjacent key frames is proposed,and the IMU data is used to achieve the first rotation and then indexing.The method realizes point cloud matching and improves the accuracy of point cloud matching between key frames;in view of the problem that the current use of motion thresholdbased methods to construct factor graphs to optimize variable nodes cann't fully utilize non-key frame point cloud information,the fixed threshold is selected and the key frame algorithm is improved.When the two sensor pose estimation results have large deviations,factor nodes are inserted in the factor graph and optimized,which improves the flexibility and accuracy of the algorithm.Finally,an autonomous navigation algorithm for unmanned vehicles based on inertial/lidar is developed under the ROS platform,and the algorithm is verified through outdoor experiments and simulation experiments.The experimental results show that the autonomous navigation algorithm can meets the real-time requirements of the system in both outdoor environment and KITTI data set,and has high navigation accuracy at the same time.