Discrete Choice Modelling for Traffic Densities with Lane-Change Behaviour

This paper investigates the modelling for traffic densities with lane-change behaviour using the information provided by loop detectors. The existing studies on traffic density estimation for multi-lane roadways mainly focus on the scenario where either vehicles’ lane-change manoeuvres are not common or the lane-change pattern is time-invariant. This research, however, takes into consideration the time-varying nature of drivers’ lane-change manoeuvres, and models the lane-change probabilities using a number of discrete choice models. These lane-change models are then embedded into a state space model to capture the dynamics of traffic flow. The extended Kalman filter is used to update the estimated traffic densities of multi-lane motorways. A numerical study is carried out to investigate the performance of the developed approach.