Lane detection is crucial information for driving autonomy. To build a safe and robust lane detection system, 3D LiDAR based lane detection, capa-ble of detecting all direction and working in all lighting condition, is an ideal sensor redundancy in addition to camera-based lane detection. The LiDAR can detect the intensity of road surface points, so the lane mark will appear as high-intensity point segment in each scan layer. The regrouping of clusters into lanes is a challenging task due to the discontinuity of road lane, the variety of road di-rection and configuration, and the presence of other road marks beside lane marks. A method has been proposed to incorporate road edge detection to pre-dict local road geometry model which assists mark-segment regrouping into lane and unrelated mark filtering. Experiments show the method is efficient, and can run in a real-time environment.
Keywords: Lane detection, 3D-LiDAR, Intensity, Road geometry
- IA-07-0017 (518K)