Vehicle self-localization in urban canyon using 3D map based GPS positioning and vehicle sensors

Precise and robust vehicle localization in the urban canyon is a new challenge arising in the autonomous driving and driver assistance systems. Sensor integration is proposed to realize this target in his paper. Global Positioning System (GPS) has been proven itself reliable for accurate vehicle self-localization in the open sky scenario. However, it suffers from the effect of multipath and Non-Line-Of-Sigh (NLOS) propagation in urban canyon. The paper proposes to estimate vehicle position by using 3-dimensional (3D) map and ray-racing method in order to overcome the problems in urban canyon. The proposed positioning method distributes numbers of positioning candidates around of reference positioning, and then the weighing of the position candidates are evaluated based on the similarity between the simulated pseudorange and the observed pseudorange. In his way, the additional 3D map information is used to reduce the effect of multipath and NLOS. Moreover, the information from vehicle sensors, including motion sensor and rotation sensor, are integrated with he GPS positioning result in a Kalman filer framework. The integration no only smoothens the trajectory of vehicle, but also reduces the positioning error. The experimental results demonstrate the accuracy of our proposed method and is feasibility for autonomous driving.