A Cooperative SLAM Method Of LiDAR And Millimeter-wave Radar In Complex Environments

Unmanned platforms can control and detect the surrounding environment by carrying various sensor devices,interacting with the environment to perform given tasks,and playing an important role in fields such as intelligence acquisition,surveillance reconnaissance,terrain survey,and forest fire.SLAM is a major component of unmanned platforms,which can provide the ego-state(such as pose and velocity)and the map of the surrounding environment for unmanned platforms,helping them to complete tasks better.Aiming at the problem of the complex and variable environment when unmanned platforms perform tasks,this thesis mainly carries out research on methods such as Li DAR SLAM,millimeter-wave radar SLAM,and cooperative SLAM of Li DAR and millimeter-wave radar in the complex environment.The main work is as follows:1.For the Li DAR SLAM problem,the working principle of Li DAR is analyzed and the corresponding observation model is established.Then the LOAM algorithm is studied.This algorithm extracts geometric features from the laser point cloud first,and then realizes SLAM based on geometric features by a two-step registration method.2.For the millimeter-wave radar SLAM problem,the working principles of Doppler millimeter-wave radar and scanning millimeter-wave radar are analyzed and the corresponding observation models are established.Then a SLAM method based on Doppler millimeter-wave radar and IMU is proposed,which realizes the Doppler millimeter-wave radar SLAM by fusing Doppler millimeter-wave radar and IMU;a SLAM method based on scanning millimeter-wave radar is studied,which uses steered BRIEF descriptor to perform feature association of radar images to realize the scanning millimeter-wave radar SLAM.3.For the cooperative SLAM problem of Li DAR and millimeter-wave radar in complex environments,the influence of complex environments on millimeter-wave radar and Li DAR are first analyzed,and a method of removing false targets from the laser point cloud in the complex environment is studied,then the cooperative SLAM method fusing laser point cloud and millimeter-wave point cloud information through iterative error state Kalman filter is proposed.