An Improved Simultaneous Localization and Mapping Method Base on LeGO-LOAM and Motion Compensation

Abstract

With the rapid development of mobile robot, the premise of all decisions and planning is to perceive the surrounding environment, especially in complex environments such as wild and mountainous areas. The mainstream Simultaneous Localization and Mapping (SLAM) algorithm Lightweight and Ground-Optimized Lidar Odometry and Mapping (LeGO-LOAM) can be well adapted to this complex environment, but it does not take into account the motion compensation of the point cloud, which leads to a decrease in perception accuracy. LeGO-LOAM adopts the way of feature point and feature point matching for posture estimation. Due to the vertical launch angle of every scanning laser radar being fixed, the radar motion will lead to distortion between the matching feature points, which can cause incorrect pose estimation. Therefore, based on LeGO-LOAM, this paper proposes a motion compensation algorithm (LeGO-LOAM-MC). Compared with the current mainstream of laser slam algorithm LeGO-LOAM, the result shows LeGO-LOAM-MC has a smoother mapping effect, smaller path drift, the maximum error is reduced by 29.1%, the mean error is reduced by 31.0%, the median error is reduced by 31.3%, the standard deviation is reduced by 37.0%, the root mean square error is reduced by 32.1%, and the sum of squares due to error is reduced by 53.9%. The experimental results show the superior performance of the proposed algorithm.