GNSS/INS/On-board Camera Integration for Vehicle Self-Localization in Urban Canyon

This paper presents a precise vehicle self-localization system for autonomous driving. The developed system integrates multiple on-board passive sensors, Global Navigation Satellite System (GNSS), Inertial Navigation system (INS) and on-board monocular camera, in order to achieve lane-level localization performance in urban environment. GNSS based positioning technique suffers from the effects of multipath and Non-Line-Of-Sight (NLOS) propagation in urban canyon. The positioning error also affects the performance in the integrated GNSS/INS system. In the other side, the lane-marking on road surface provides the important visual source of information for driving. This paper proposes to detect the occupied lane of vehicle using on-board cameras. The lane detection result is integrated with GNSS/INS system in order to improve the positioning error. The experiment results demonstrate that the proposed method can provide 90% correctness with respect to the occupied lane.