Road traffic queue length estimation with artificial intelligence (AI) methods

Sustainable monitoring of traffic has always been a significant problem for engineers, queue length being one of the most important metrics required for the performance assessment of signalized intersections. The authors of the present study propose a novel approach to estimating cycle-by-cycle queue lengths at a given signalized intersection. Focusing on the examination of shock wave phenomena and the traffic model, this study first elucidates the definitions and assumptions it employs. Subsequently, it delves into the creation of the queuing model, alongside the utilization of a machine-learning (ML) based Kalman Filter (KF) algorithm for estimation. The information contained within the output files is visualized on distinct graphs, along with the velocities at various time intervals derived from virtual simulations involving a queue of 12 vehicles. This graphical representation serves as a conclusive validation, demonstrating a strong correlation between the simulation and the estimation achieved through the KF approach. The method presented yielded dependable and resilient estimates for the simulated queue lengths, even in the presence of noisy measurements.