Estimation of Traffic Densities for Multilane Roadways Using a Markov Model Approach

Inductive loop detectors are widely deployed in strategic roadway networks. This paper investigates recursive estimation of traffic densities using the information provided by loop detectors. The existing studies for multilane roadways mainly focus on the scenario where vehicles' lane-change movements are not common and can be ignored. This paper, however, takes into consideration lane-change effect in traffic modeling and incorporates a Markov chain into the state space model to describe the lane-change behavior. We update the traffic density estimate using the Kalman filter. To avoid the approximation due to the linearization of the nonlinear observation equation in the extended Kalman filter, we have considered a suitable transformation. Numerical studies were carried out to investigate the performance of the developed approach. It is shown that it outperforms the existing methods.