Design of traffic signal timing schemes for intersections based on vehicle spatio-temporal trajectories

In scenarios where vehicle spatio-temporal trajectory data is available, this study proposes a real-time optimization scheme for single-point signal control based on the trajectory data of incoming vehicles, taking into account drivers' reliance on signal phase transition countdowns. The aim is to address the mismatch between signal control schemes and the randomness of traffic flow arrivals. Firstly, the concept of stage division is introduced, and a two-stage cyclic real-time control scheme is constructed. Secondly, vehicle's passing during the green phase and vehicle’s queuing during the red phase are predicted based on trajectory data. To balance vehicle throughput and delay, a method is proposed to determine the remaining green time during the variable stage of the scheme. Finally, using an actual intersection as a case study, a simulation experiment platform is built with SUMO and Python to compare the proposed signal control method with the traditional Webster timing method. Simulation results demonstrate that the proposed signal control scheme outperforms the Webster timing method in terms of average speed and delay indicators, regardless of peak or off-peak periods. These findings suggest that vehicle spatio-temporal trajectory data can facilitate the development of signal timing schemes that better align with the actual traffic conditions at intersections, offering a new approach for achieving refined control of signalized intersections.