Based On Local Map Joint Optimization LiDAR-IMU Tightly Coupled SLAM Algorithm Research

With the progress of science and technology and social development,mobile robots are more and more widely used in various fields,and play an important role in power inspection.The technology of robot location and mapping(SLAM)is the key technology which plays a decisive role in the development of intelligent mobile robot.Especially when mobile robots are more and more used in complex and changeable working environment,the single sensor SLAM technology has been unable to meet the application requirements,and the multi-source sensor fusion based slam technology has become a new development direction.Combining the development trend of multi?source sensor fusion and the development status of laser SLAM,this paper aims to improve the intelligence of mobile robot's environmental perception and its positioning accuracy in complex environment.Starting from lidar and IMU sensors,this paper studies the algorithm of lidar-imu tightly coupled SLAM.The main research contents can be summarized as follows:Firstly,the nonlinear optimization of IMU pose data based on residual is realized.The IMU noise model is established,and the IMU motion estimation and LiDAR point cloud datas alignment are realized by combining the kinematic model.The preintegration of IMU measurement is realized,and the local optimization of IMU pose data is carried out in the nonlinear optimization framework.After that,the LiDAR point cloud preprocessing based on IMU data is realized,and the point cloud registration is based on the feature.Using IMU pose detection data,the distortion caused by high-speed motion in LiDAR point cloud is removed by linear interpolation.Comb and denoise point clouds to reduce the number of point clouds.The edge and plane features are extracted,and then the whole point cloud registration is realized.Then,the off-line external parameter calibration and time synchronization of lidar and IMU are carried out by using the non-linear optimization method to achieve the time and space alignment.The optimization residual of local lidar is established by sliding window,and the joint optimization framework is realized by combining the prior information.Through the optimization of position and attitude map and loop detection,the optimization of trajectory map is realized,and the cumulative error of odometer is corrected.Finally,an experiment is designed to verify the effectiveness of the proposed method.Compared with lgeo_loam,it has smaller cumulative error and higher accuracy.