Research on point cloud matching of lidar based on odometer

Abstract

Multisensor information fusion of mobile devices is the basis of the research of mobile location. Odometer and laser point cloud are the main methods to determine the pose in existing positioning techniques. But the initial pose uncertainty, complex iterative process and longer time consuming problems will cause the positioning accuracy greatly reduced, especially after a long distance movement. Therefore, a general odometer-assisted method is proposed for the wheeled mobile platform based on laser point cloud matching. In the approximate motion hypothesis of wheeled mobile platform based on the path curve, rough pose estimation based on the characteristics of the odometer first mobile platform, and this attitude for the initial attitude laser point cloud matching, with the attitude as point cloud matching iteration begins. Experimental results show that the odometer in pose positioning way, effectively reduce the accumulative error of point cloud matching, improves the accuracy of pose determination; also increased the time spent in iteration, improves the work efficiency of the device. In the assumption that the wheeled mobile platform movement path approximates the arc, according to the work characteristic of odometer. Firstly, a rough pose estimate of the mobile platform is presented, and the initial pose of the laser point cloud is taken as the initial value of the matching iteration of the point cloud. Experimental results show that the odometer in pose positioning effectively reduce the accumulative error of point cloud matching and improves the accuracy of pose determination. At the same time, it also reduce the iteration time cost and improve the work efficiency of the device.