Real-time travel-time prediction method applying multiple traffic observations

Various methods have been developed to predict automobile travel time, but they are often unreliable, especially when the travel time varies significantly during the transition between free flow and congested flow. This paper proposes a real-time travel-time prediction method. We apply a macroscopic traffic flow model with predicted boundary conditions and modify the scheme to calculate the traffic states to reflect the latest traffic conditions on a real-time basis. Our method uses traffic data from multiple observation systems, which is a crucial component for real-time application of the macroscopic traffic flow model that has not been previously applied to traffic flow models. The analysis of real traffic data collected from a section of the Korean Kyungbu Expressway shows that the proposed method outperforms other prediction methods, particularly during the transition between free flow and congested flow.