In this paper, a hybrid positioning method is proposed to estimate position of a vehicle continuously in urban canyon by combining a single frequency GPS receiver and a low cost monocular vision sensor. As widely known, GNSS cannot provide sufficient positioning accuracy and continuity because insufficient number of visible satellites in urban canyon. By utilizing the vision sensor, the proposed method estimates direction and state of the vehicle when driving in urban canyon. The proposed method can estimate trajectory of the vehicle by combining information of the vehicle and GNSS satellite signals. Therefore it is possible to overcome shortcoming of GNSS and improve positioning accuracy and continuity. The proposed method does not need extra information such as map and geographic information. Only two satellite signals and road images are required for the application of the proposed method. By an experiment result utilizing field-collected measurements, feasibility of the proposed method is demonstrated.

Keywords: GNSS, vision sensor, hybrid positioning method