Urban Road Travel Time Prediction Based on Taxi GPS Data

An approach to estimate the urban road travel time by using probe vehicle data collected through taxi GPS was proposed, and GPS data correction method of probe vehicle was introduced. According to the poor adaptive ability of traditional filter algorithm in the estimation for traffic state and travel time with Kalman filter, an improved adaptive Kalman filtering method was proposed. The method overcame the shortcomings that the traditional filter couldn't real-time track the change of environment. It could adapt to the dynamic changes of traffic state to realize the optimization estimation. The method was tested on urban roads in Guangzhou, and there was a slight difference between the prediction result and that of actual observation in free traffic flow state and stable flow, and the relative error was under 10% in traffic congested state. The results show that the traffic state estimation model had better tracking ability than conventional Kalman filter.