关键词: 车辆导航, 地图匹配, 短时预测, 证据理论, 信息融合

Abstract: Efficient and reliable map matching algorithms are essential for vehicle navigation systems, while most existing solutions cannot provide trustworthy outputs when the situation is ambiguous (such as at road intersections). In order to improve the precision of map matching, a new map matching algorithm based short-term prediction was proposed. Firstly, the algorithm employed the history positioning information to set up the model of short-term prediction. Accordingly, the future positions would be obtained after the current matching time; secondly, the distance comparability between vehicle and route was defined by the modified average distance, and it replaced the projecting distance between current matching position and route; finally, the Dempster-Shafer evidence was adopted to fuse the modified average distance and direction information between vehicle and route. It could effectively expand the credibility differences of the candidate routes and enhance the robustness of the algorithm. The results of simulation and experiments demonstrate the better efficiency and reliability of the estimates even for ambiguous environment.

Key words: vehicle navigation, map matching, short-term prediction, evidence theory, information fusion