On Map-Matching Algorithm And Its Application In Vehicle Navigation System

Map-Matching,integrate the vehicle positioning data with digital road network,is an important positioning technique in the vehicle navigation system.To improve the accuracy of map-matching,this paper,using different algorithms and model combinations,focuses on the following aspects for in-depth research:For improving map-matching accuracy in cities having dense overpasses,we firstly proposed a map-matching algorithm based on cloud model.This approach,assisted altitude data,established a map-matching model by establishing cloud rules and reasoning with uncertainty based on cloud model.The cloud model integrated fuzziness with randomness of qualitative concept so as to overcome the subjective randomness in fuzzy membership grade when being determined.To improve the disadvantage in the inadequate information of map-matching algorithm using current positioning point only,this paper proposed an intelligent map-matching algorithm utilizing current positioning point as well as some history positioning points.In this algorithm,a distance between two trajectory curves was defined by a average Fr(?)chet distance measure,and then a map-matching algorithm based on cloud controller was proposed so that the credibility,P,as a evaluating index of map-matching was outputted.The proposed algorithm could provide not only warning information to user when mismatching occurred,but also a means of fast recovery from failure.The simulation test demonstrated that the general accuracy of the proposed algorithm was superior to that of matching only using the current positioning point.In order to enhance the self-adaptive of matching algorithm,an adaptive-fuzzy-network based on C-measure map-matching algorithm and its advantages were briefly summarized firstly,in which the C-measure was defined to represent the certainty of the car's existence on the corresponding road.But,as this algorithm emphasizes on current positioning data only,the matching accuracy decreases in complicated road network due to the lack of data.In order to improve precision of vehicle tracking system,a strategy was proposed.This strategy employed history positioning information to overcome the disadvantage of the original algorithm in information insufficiency,and the distance between two history trajectory curves was defined by an average Fr(?)chet distance measure to implement curves matching instead of point matching.Owing to increase historic information input variable in the fuzzy network,the number of fuzzy reasoning rules was increased,and operating efficiency of the fuzzy network was reduced.For this reason,a scheme to simplify reasoning rules and to enhance the efficiency was proposed by using hierarchical fuzzy control technique.Additionally,the learning algorithm was updated to support the algorithm. The experimental results demonstrate the effectiveness of this proposed algorithm.Finally,a novel map-matching algorithm based on multi-criteria fusion using belief theory was proposed,which provided a synthetic evaluation for the results of road selection in road matching.Firstly,a proximity criterion and an angular criterion were established on current information in vehicle navigation,and a historical trajectory proximity criterion was put forward according to historical information.The road selection strategy then was employed based on multi-criteria fusion by using D-S evidence theory.Finally,the accuracy of matching results was improved by taking account of connection criterion.Simulation results demonstrate that the proposed algorithm is effective and can recognize the road on which the vehicle is running in simple and complicated road network.