A 3D LiDAR Data-Based Dedicated Road Boundary Detection Algorithm for Autonomous Vehicles

Abstract

Effectively detecting road boundaries in real time is critical to the applications of autonomous vehicles, such as vehicle localization, path planning, and environmental understanding. To precisely extract the irregular road boundaries or those blocked by obstructions on the road from the 3D LiDAR data, a dedicated algorithm consisting of four steps is proposed in this paper. The steps are as follows. First, the 3D LiDAR data is pre-processed, employing the vehicle position and attitude information, and many noise points are deleted. Second, the ground points are quickly separated from the pre-processed point cloud data to reduce the disturbance from the obstacles on the road; this greatly decreases the size of the points cloud to be processed. Third, the candidate points of the road boundaries are searched along the predicted trajectory of the autonomous vehicle and filtered using the unique features of the boundary points. Last, a spline fit model is applied to smoothen the road boundaries. An experiment to test the performance of the proposed algorithm was conducted on the “Xinda” autonomous vehicle under various road scenarios. The experimental results show that the average accuracy of the proposed algorithm exceeds 93%, and its average processing time is approximately 36.5 ms/frame, which outperforms most of the state-of-the-art methods. This indicates that the proposed algorithm can robustly extract the road boundary in real time even if there are many obstacles on the road. This algorithm has been tested on “Xinda” autonomous vehicle for over 1000 kilometers, and its performance was always stable.