Comparison of Bi-level and Nonlinear Optimization for Urban Traffic Control

Abstract

The goal of this research is to improve the traffic behavior at crossroads of the urban network by decreasing the waiting vehicles in the network. This goal is achieved through bi-level optimization, which has significant advantages compared to classical optimization. The added value in this research is the analytical formalization of the optimization problem. The problem gives superiority to the values of the outgoing flows in comparison with the incoming traffic flow in each direction of the network, which results in the decrease of waiting vehicles. The research applies control rules of predictive control at each traffic light cycle by the bi-level optimization methodology. This enables additional adaptation of the green light and cycle durations to the traffic flows. The bi-level solutions are numerically evaluated and compared with the classical (one-criterion) optimization. The comparison demonstrates the advantages of bi-level optimization.