Design of a supervisory controller for Cooperative Intersection Control using Model Predictive Control

Abstract

Abstract The Cooperative Intersection Control (CIC) methodology ensures a safe, smooth traffic flow through an automated intersection by means of virtual platooning. In this paper, we address the design of a centralised supervisory controller for CIC which optimises the crossing sequence of vehicles. We propose an approach in which the intersection as a whole is modelled as a hybrid system, which evolves in both continuous-time and in discrete-time. This hybrid system model resembles a queueing system, and relates the entry of vehicles into the intersection to a measure of the delay of their travel through the intersection. We design a supervisory controller using Model Predictive Control (MPC), which aims to minimise the vehicles’ average delay by controlling their access to the intersection. A simulation study based on real-life data demonstrates the effectiveness of the MPC approach compared to a first-come-first-served (FCFS) policy and a conventional traffic light controller. This study shows that MPC achieves a faster transient response and a lower average delay, thereby increasing the throughput of the intersection.