A Point Cloud-Based Robust Road Curb Detection and Tracking Method

Abstract

Road curb detection is essential for autonomous vehicles to locate themselves and make a rational decision, especially under road discontinuities, obstacle occlusions, and curved road scenarios. However, an effective and systematic solution to this problem has remained elusive. In this paper, a robust 3D-LiDAR-based method for road curb detection and tracking in a structured environment is proposed. The proposed method consists of four main stages: 1) a multi-feature based method is applied to extract candidate points; 2) a density-based clustering method is proposed for classifying left and right candidate points; 3) a candidate points filter (including distance filter and RANSAC filter) is proposed to remove false points; and 4) a least-square algorithm is used to obtain road curb curve and the amplitude-limiting Kalman filter is deployed to prevent false detection and miss detection. The comprehensive experiment evaluations show that the proposed method can deal with straight and curved road without being influenced by surrounding obstacles.