Laser Scan Matching based on Normal Distribution Transform using Efficient Point Cloud Segmentation

The Normal Distributions Transform (NDT) is currently a very practical laser scan matching algorithm. In this paper, a NDT variant algorithm is designed for SLAM which contains a ground point remover and a effective point cloud data segmentation. Its primary goal isto improve the robustness oftheNDTalgorithm. By extracting points belonging to the ground and segmenting 3D scan data into different objects, the proposed method solve the problem that the cost function of the standard NDT is not continuous enough. Simultaneously, the computation speed of the algorithm increased significantly. Final ly, experimental results are presented to illustrate the effectiveness of the proposed algorithm.