Decentralized Intersection Management of Autonomous Vehicles Using Nonlinear MPC

The rapid population growth and increase in vehicle numbers over the last few decades have caused traffic congestion worldwide, as intersections significantly impact the efficiency of traffic networks in urban areas, this paper focuses on intelligent traffic management at intersections by integrating the technologies of Intelligent Transportation Systems (ITS) and Autonomous Vehicles (AV). In this framework, a physical model represents each vehicle taking into account the dynamic limits. A decentralized Nonlinear Model Predictive Control (NMPC) is proposed to help coordinate the traffic flow of the AV at the intersection. The controller solves a quadratic cost function for the vehicle to ensure a smooth trajectory and minimum energy consumption, while avoiding collisions that is guaranteed using linear constraints and two developed priority assignment methods, Predicted Arrival Time (PAT) and First-Come First-Served (FCFS). The two methods are tested using a developed simulation environment on MATLAB/Simulink and the results are compared, the predicted arrival time method outperformed the FCFS method.