Correction of Vehicle Positioning Error Using 3D-map-GNSS and Vision-based Road Marking Detection

Accurate and robust vehicle self-localization in the urban environment is a new challenge arising in the autonomous driving. GNSS positioning technique suffers from the effects of multipath and Non-Line-Of-Sight (NLOS) propagation in urban canyon. This paper proposes to employ an innovative GNSS positioning technique with the aid of 3D building map, to mitigate the error caused by multipath and NLOS. In addition, the road markings on road surface provide the significant visual information for driving, which can also be used for localization. Based on the lane marking detection, the lane keeping and changing behavior can be recognized and used for positioning. This paper integrates 3D-map-GNSS with vision-based road marking detection and 2D Map to reduce the vehicle positioning error. The experiment results demonstrate that the proposed method can provide sub-meter accuracy with respect to positioning error mean.