Point Cloud Segmentation Algorithm for Flatbed Trucks Based on LIDAR

To optimize the efficiency and accuracy of the loading process in the logistics field, and to address the issues of insufficient robustness and under-segmentation of truck point cloud segmentation algorithms in complex loading and unloading environments, a point cloud segmentation algorithm for flatbed trucks based on LIDAR is proposed. First, an improved adaptive voxel filtering algorithm is used to dynamically adjust the amount of point cloud data according to the local density of the point cloud, achieving the purpose of down-sampling. Secondly, a combination of pass-through filtering, radius filtering, and statistical filtering is used to remove noise of different sizes and types. Furthermore, a two-stage truck point cloud segmentation algorithm combining an improved normal-based segmentation algorithm and Euclidean clustering is designed. In the first stage, an improved normal-based segmentation algorithm is used, combined with PCA principal direction transformation, multi-feature fusion, and a weighted scoring mechanism to achieve preliminary orthogonal three-direction segmentation of the truck. In the second stage, the Euclidean clustering algorithm is used for fine segmentation, obtaining the segmentation results of each region of the truck. Experimental results show that, compared to other segmentation algorithms, this method can effectively extract and distinguish each region of the flatbed truck, demonstrating high segmentation accuracy.