Global Sensitivity Analysis of the Optimal Capacity of Intersections

While the problem of optimizing traffic light phases dates back to the fifties, in the daily practice the optimal solutions are often still determined by means of deterministic approaches that assume to know exactly the incoming traffic flows and the other intersection parameters. In this framework, the problem of parameter uncertainty is normally neglected, and the flow variability is only taken into account by the responsive traffic light plans. In this paper, a hybrid model of traffic light dynamics, and a global sensitivity analysis approach are proposed with the aim of providing a methodology for evaluating the performance robustness of different traffic light optimization approaches with respect to parameters uncertainty and variability. To this end, two well known solution approaches are compared and discussed by means of a real world case study.