Research On Technology For Getting Traffic States Based On Map Matching

In the present, traditional immovable detector like loop detectors, ultrasonic detectors, infrared detectors and video detectors are the main data source of ITS in traffic state. Inherent defects such as limited range and equipment failure of those detectors resulted in the lack of some transport information. It's difficult for traditional detectors to reflect the traffic state of urban roads completely and accurately. GPS is a better way to resolve those problems.The main contents of this paper contains analyzing the sources of errors in GPS data and make corrections on those errors by Calculus of Differences method to match GPS data to map. Calculus of differences method improved the matching accuracy effectively. Method of Grid Indexing could greatly improve the matching speed and maintain the real-time feature of map matching. Inherent defects of traditional detectors as limited range make it difficult to reflect the state of the regional transport accurately. Location movable feature of GPS data make it more suitable for reflecting the status of regional transportation. I used the Circle Split method in this paper to match GPS positions to their traffic areas, the speed curve by calculating is consistent with a normal speed curve. In this paper, I used Travel Time method to get the traffic states of roads from GPS data after map matching. The results show that it's closer to the traffic state curve obtained from loop detectors. Besides, I did some basic work in this paper as defining the data model of road network. On one hand this model was designed with the realization of the above algorithm requires. It's highly compatible with Calculus of Differences method, Circle Split method and Travel Time method. On the other hand this model is more flexible to meet the future expansion of more complex functions.This paper based on the program subject'Olympic electric vehicle demonstration and technical support' which was supported by the national science and technology. There're 28 thousand float cars in this project which is a great number and it has a good representation to ensure the accuracy of the algorithm results in this paper.