Real-Time Freeway Traffic State Estimation and Incident Detection based on Extended Kalman Filter: An Overview

Recent advance in real-time freeway traffic state estimation is reviewed in this paper, with a particular focus on a general approach to traffic state estimation based on joint state and parameter estimation and another focus on the estimation performance in large-scale field applications. A mathematical model is first presented, including a validated macroscopic traffic flow model and a measurement model. The traffic state estimator is designed on the basis of extended Kalman filtering. The estimator's performance in tracking capability and automatic incident detection is then carefully examined via real-data tests or field applications, under various conditions involving large-scale networks, sparse measurements, infrastructure heterogeneity, changes of environmental conditions, traffic incidents, detector faults, and measurement inaccuracy. The paper is closed with discussions and conclusions. This paper overviews the state-of-the-art research on real-time freeway traffic state estimation using extended Kalman filtering. Ongoing research in this area addresses several extensions of the work reported, including: Comparison with other filtering methods (particle or unscented filters); decentralisation of the filtering structure; state estimation in urban road networks; new issues arising from the increasing presence of connected and even automated vehicles; and more.