A method for classifying the traffic states of signalized intersection using probe travel time

Abstract

This study aims to develop a next-generation traffic signal control system, which adapts to the observed real-time traffic states obtained by probe observations. In this paper, 4 types of traffic states, for which different control policies should be applied for efficient signal control, are defined. Then, a method is proposed to classify the intersection traffic states into the 4 types based on the real-time probe travel time. Specifically, the probe travel time is utilized to determine whether or not the discharge flow rate of the intersection is decreased and whether or not congestion appears at each signal phase; then the current traffic states are classified into one of the 4 states based on these results. The accuracy of the detection of discharge flow rate reduction is verified using actual observation data, and differences in the accuracy are observed at different intersections. In the future study, factors that affect the classification accuracy will be examined to improve the accuracy.