Iterative Learning Control for Highway Traffic Control

Abstract

In the modern world, congestion on the highway has become a significant social problem due to the increasing number of vehicles. This leads to considerable waste of time and pollution. On the other hand, service stations play an important role in affecting the congestion on highways. In this project, we aim to study the effect of controlling the outflow of the service station on the overall congestion on the highway. Moreover, the vehicle flows on the highway traffic have repetitive patterns from day to day, week to week, which motivates us to apply iterative learning control (ILC) to the service stations. Based on the proposed Cell Transmission Model with service station (CTM-s), we formulate the nominal model predictive control (MPC) problem. However, due to uncertainties on drivers’ intentions, several parameters of the nominal MPC are unknown and can only be roughly estimated. We implement ILC to deal with these uncertainties by leveraging past data. After comparing the control effect of MPC (using estimated parameters) and ILC on the macroscopic CTM-s model, we validate the results on the high-fidelity microscopic traffic simulator Simulation of Urban MObility (SUMO).