Real-time extraction method of road boundary based on three-dimensional lidar

The problem of road boundary extraction in the structured road environment where there are discontinuities, occlusions and dynamic vehicles is critical to autonomous driving, but a solution to this problem has remained elusive. In this paper, a new straight-line method for real-time extraction of road boundary is proposed. Three-dimensional (3D) lidar is used to recognize road boundary with random sampling consensus algorithm (RANSAC) and certain filters. First of all, the boundary point cloud data are extracted with the method by calculating curvature change and height difference for each laser layer, then RANSAC algorithm is used to fit the extracted boundary points, then get the straight-line equation of the road edge. Finally, the Amplitude-Limiting filter and the Kalman filter are applied synthetically to the obtained boundary equation, remove unreasonable line equation and smooth the boundary line jitter. The real vehicle test proves that this algorithm can accurately and robustly perceive the curb boundary under a variety of road conditions mentioned above. The operation time of the on-board industrial computer is 26.7ms which fully meets the real-time requirements of unmanned vehicle.