Robust vehicle tracking fusing radar and vision

Many driver assistance systems are based on vehicle detection and tracking including adaptive cruise control, collision warning and fully autonomous driving. A large detection range is required, especially while driving at higher speeds on highways. A reliable and precise detection is needed even under adverse weather conditions. In this paper we present a fusion approach combining radar and monocular image processing. The approach enables one to track vehicles up to a distance of 130 m and to assign them reliably to specific lanes.