Real-time Monitoring of Dynamic Traffic States by State-Space Model

We propose real-time traffic state estimation using a state-space model that takes account of variability in the fundamental diagram (FD) and sensing data. In free flow situations, for instance, the FD regulating driving behavior may vary among drivers who possess differing characters. In addition, FD is affected by external factors such as interactions with pedestrians and vehicles. Variational theory (VT) was used as the system model, and measurement data were taken from probe vehicles and traffic detectors. VT is a static model, making real time estimation of traffic state changes difficult. We applied VT to a state space model. Our proposal showed better agreement between simulated and benchmark traffic states than deterministic VT. For model validation, we applied the model to Komazawa Street in Tokyo, Japan. Copyright (C) 2017 The Authors. Published by Elsevier B.V.