测绘学报

• 学术论文 • 上一篇    

基于路径预测的不确定性推理组合地图匹配算法

唐进君,刘芳   

  1. 内蒙古农业大学
  • 收稿日期:2009-09-23 修回日期:2010-04-03 出版日期:2010-10-25 发布日期:2010-10-25
  • 通讯作者: 唐进君

A Driver Route Prediction Based Map-matching Algorithm Integrating Uncertain Reasoning

  • Received:2009-09-23 Revised:2010-04-03 Online:2010-10-25 Published:2010-10-25

摘要: 地图匹配是车辆导航系统中的重要组成部分。本文在简要介绍相关地图匹配算法和总结前期研究的基础上,提出了一种新的基于预测的不确定性推理组合地图匹配算法。该方法首先利用云模型这样一种处理定性与定量间不确定转换的工具,通过对当前信息的不确定性推理,计算定位点相对于候选道路的匹配可信度。之后,利用隐马尔科夫模型预测驾驶员出行路径及目的地,通过对模型中的观察值函数重新设定,将路径预测信息与当前地图匹配算法结合,构建组合地图匹配算法。同时,采用学习算法更新模型中的信息。最后,通过仿真实例证明了该组合算法的有效性,结果表明:该方法能够预测驾驶员行驶路径,起到预匹配的作用,提高了地图匹配算法的精度与实时性。

Abstract: Map-Matching is one of the important parts of the vehicular navigation system. This paper provided a brief introduction of various map matching algorithm firstly, a novel integrated algorithm combining route prediction with uncertain reasoning was proposed based on previous researching. In the proposed algorithm, the cloud model, a powerful tool to perform uncertain converting between numerical quantitative analysis to conceptual qualitative analysis, was firstly used to calculate the credibility of positioning point to candidate roads by uncertain reference with current locating data. Then, a Hidden Markov Model (HMM) was built to predict the driver’s routes and destinations. Through redesigning observation function in HMM, an integrated map matching algorithm was established by combining route predicting information with current matching approach. Additionally, the learning algorithm was carried out to support the algorithm and update the information. Finally, the experimental results demonstrate the effectiveness of the proposed algorithm: it can predict a driver’s routs to improve the accuracy and real-time characteristic of map matching algorithm according to the utility of pre-matching.