Vehicle Detection Framework Based on LiDAR for Autonoumous Driving

Abstract

Stable and efficient environmental perception in autonomous driving is an important prerequisite for path planning and behavior prediction. This paper proposes a vehicle identification framework based on LiDAR. After using the ground segmentation algorithm based on Gaussian process regression to complete high-quality ground segmentation, the adaptively density-based spatial clustering of applications with noise (A-DBSCAN) algorithm is used to cluster the remaining obstacle point clouds, and then the improved L-Shape algorithm is used for bounding box fitting. The experimental results based on the KITTI data set show that the framework has good stability under simple working conditions.