Dynamic route guidance using GPS equipped taxi data

Dynamic route guidance is one of the focuses of recent researches on ITS, but its application is limited by the locations and numbers of traffic detectors. This paper advanced a set of methods for traffic route guidance using GPS equipped taxi (probe vehicle) data. Firstly, the paper proposed a map-matching method, eliminated ineffective data by observing the status of taximeters and calculated link time cost of each probe vehicle. Secondly, the paper developed a dynamic optimal route choice arithmetic based on adjusted Q-learning, which can get the expect route with links' weights model unknown. Thirdly, rout choice table was established for route guidance. Finally, the results of simulation experiment based on the electronic map of Guangzhou city indicate that the methods advanced in this paper are quite effective and practical.