Tightly coupled LiDAR-inertia odometry based on multi-class feature point matching
Aiming at the problems of low precision of laser odometer due to motion distortion of LiDAR data and sparse ground sampling data,a tightly coupled IMU odometer method based on multi-class feature point matc-hing is proposed in this paper.In order to improve the quality of LiDAR data,it firstly starts with the original point cloud data and then uses IMU data to perform linear interpolation to correct the distorted point cloud in each frame of LiDAR data.Secondly,after distortion correction,it performs a 2D grid projection on the point cloud.According to the average minimum height of each grid and its adjacent grids,the point cloud in the grid is divided into ground points and non-ground points,using a dual threshold.Then,the non-ground points are further divided to obtain multi-class feature points according to linearity,flatness,curvature and other local fea-tures.Thirdly,it models the tight coupling of IMU for multi-class feature point matching.Considering that the original LiDAR observation error cannot provide high-precision gravity vector estimation,it introduces IMU state estimation,builds odometer constraint error function,and makes a further constraint on the estimation of gravity vector.Thus,the precision of laser odometer is improved effectively.Finally,an IMU tightly coupled laser o-dometer based on multi-class feature point matching is designed based on LeGO-LOAM framework,and the veri-fication system is completed.Experimental results show that this method can effectively suppress the drift of gravity vector and improve the precision of laser odometer.
multi-class feature pointsfeature point matchinglaser odometerIMU tight couplingSLAM
万方数据
维普
