Fast Positioning Method of Truck Compartment Based on Plane Segmentation

Abstract

To improve the loading efficiency and shorten the loading time, a method based on LIDAR point cloud segmentation for fast positioning of trucks is proposed for the scenario of level-push automatic loading, in which the truck position and attitude are random. The point cloud data are collected by LIDAR and pre-processed by traditional methods in the order of straight pass filtering, voxel filtering, and plane segmentation. Four planar regions are segmented using the region growth algorithm, and then planar fitting is performed for each of these four regions, which requires a large number of iterations and complex calculations by the traditional planar segmentation method. Focusing on these drawbacks this paper first uses the region growing algorithm to cluster and group the matching points, then combines the least squares and RANSAC algorithms for plane fitting optimization to calculate the intercept of each plane, and finally uses these intercept values to perform spatial geometry calculations to obtain the exact position and locus of the carriage. Experiments show that the method reduces the number of iterations and improves the computational efficiency compared with the traditional RANSAC algorithm, thus improving the efficiency and accuracy of the plane fitting algorithm.