A Model Predictive Control Approach for Smooth Traffic Flow of Connected Autonomous Vehicles at Signalized Junctions

In this article, we consider the problem of controlling a group of autonomous vehicles, which are maneuvering on a road, such that smooth traffic flow can happen. For this, we propose that every autonomous vehicle maneuvers with a prespecified velocity profile. The velocity profiles are determined in such a way that either none of the vehicles wait near a signalized junction or only a limited number of vehicles are accumulated when the red light is ON. To maneuver the autonomous vehicles with a pre-specified velocity profile, a control sequence is generated using a receding horizon control approach in the model-predictive control framework. The associated problems, such as the feasibility of the underlying constrained optimization problem at every stage and the convergence analysis, are also considered. The efficacy of the proposed approach is evinced with numerical simulations.