基于路网拓扑特性及先验知识的地图匹配算法

Map-matching algorithm based on road network topological information and priori-knowledge

基于最大似然估计原理,提出了地图匹配的统一数学模型.将二维地图匹配算法的估计拓展到多维估计,利用道路网络的拓扑特性及先验知识修正前述算法.通过先验知识数字化,用代价函数将基于拓扑特性和先验知识的地图匹配算法纳入统一数学模型中.通过在香港大量的车辆定位实验,及参数修正和模型改进,对前述算法进行验证.实践表明,通过综合利用全球定位系统、航位推算以及数字地图道路网络等多种信息,该地图匹配算法在大规模复杂网络以及高楼环绕的场合可以达到很好的定位效果.

Based on maximum likeness estimation, a unified math model of map-matching algorithm is presented. Then, 2-D map-matching algorithm based on normal distribution is given, and its disadvantages are discussed. Based on aforementioned research, through extending 2-D to multi-dimension, by using topological information of road network and priori-knowledge, an improved map- matching algorithm is obtained. In this algorithm, priori-knowledge is represented using cost function or likeness function. Therefore, the presented map-matching algorithm can be combined into the unified map-matching mathematic model. Many vehicle navigation tests in Hong Kong are done for parameter correction and model improvement. In the end, a vehicle-positioning test with different areas in Hong Kong is given. A conclusion can be drawn that, through integrating kinds of information like GPS （global positioning system）, dead reckoning（ DR）, road network information, the algorithm proposed can give satisfactory positioning performance in large-scale dense road network and area with many high building around.