Automatic Design of Optimal Actuated Traffic Signal Plans with Active Transit Priority

Efficient traffic signal control design is essential to provide a high level of service at intersections. Current signal control plans are increasingly complex due to advances in sensing and communication capabilities, and the introduction of advanced control functionalities, such as transit signal priority or pedestrian actuation. Most optimization programs for actuated signal plans focus on setting the values of the main design parameters, such as cycle length and maximum and minimum green times. Other design components, such as phase compositions, detector placements and priority strategies, are selected by traffic engineers based on their experience or according to general engineering guidelines. This paper presents a system and its various components that fully automates the design process and generates optimal actuated signal plans with transit signal priority functions. The system consists of two main components, a traffic simulation model that evaluates candidate signal plans and a grammatical evolution algorithm that generates these plans within the optimization process. The control design system was evaluated with a case study on an intersection in Haifa, Israel controlled by an actuated plan with transit priority to Bus Rapid Transit (BRT) vehicles. An automatically designed signal plan improved average person delay by 25% compared to the existing plan. It also outperformed, by 3%, a signal plan that kept the existing structure, and optimized its timing parameters (i.e., cycle length, minimum and maximum green time). Thus, the system can generate plans that are at least as good as human-designed ones, with lower effort, cost, and design time.