טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
M.Sc Thesis
M.Sc StudentKomarovsky Shimon
SubjectRobust Traffic Signal Control for Uncertain Road Networks
DepartmentDepartment of Civil and Environmental Engineering
Supervisor Professor Jack Haddad
Full Thesis textFull thesis text - English Version


Abstract

As congestion in road networks remains a severe and current issue, many traffic control strategies try to tackle this problem. Many of such strategies in the literature are designed based on various traffic flow models, which do not include parameter uncertainties in their intersection queuing dynamics. Hence, these control strategies \blue{might handle poorly conditions when} assumed values for the parameters vary significantly from their real values. Moreover, many utilized state feedback control approaches to develop the traffic signal controllers lack the treatment of control and state constraints directly in the design phase.


This thesis considers robust traffic signal control for uncertain urban road networks. First, parameter uncertainties are integrated into the store-and-forward (SF) model, which is utilized in this thesis to describe the queuing dynamics for traffic-signalized intersections. The SF model is unique in a sense that it simplifies traffic flow description from discrete-time modes to a continuous-time mode over a cycle, by averaging the flow throughout the whole cycle. This simplification allows the implementation of various optimization and control methods and schemes.


The SF model is first upgraded to include traffic flow uncertainties to reflect real urban traffic dynamics. Then, the uncertain SF model is utilized to design a robust feedback controller by an interpolation-based approach. In this thesis two versions of this approach are implemented: Interpolating Control (IC) and Simple Interpolating Control (SIC). This approach (i) guarantees robustness against all \blue{assumed} parameter uncertainties, (ii) handles  control and state constraints, and (iii) presents a computationally cheap solution.


Finally, numerical results for an isolated signalized intersection show a comparison between the developed interpolating controllers and other controllers in the literature. The results demonstrate the performance advantages from applying the robust interpolating controllers.