Real-time adaptive signal controller for non-lane following heterogeneous road traffic

Signalized intersections play a critical role in urban road transportation systems to reduce congestion. Managing these signal control systems at their optimal performance has been an important research focus in traffic engineering. However, various optimization techniques and algorithms developed for adaptive signal control are less effective for real-time implementation in non-lane following heterogeneous road traffic due to the following possible reasons: erroneous demand estimation, time-expensive computations, or uncertainty of getting optimal solutions. Rather, these models work efficiently for homogeneous lane-following traffic where demand predictions are accurate. This work proposes an optimization model that minimizes the total average control delay and finds the optimal green times without any explicit demand prediction for heterogeneous non-lane following road traffic. The proposed solution approach finds the optimal green times and the cycle time in real-time, and the obtained solutions are adaptive to traffic fluctuations. Simulation results show significant reduction in the average control delay and average queue length as compared to vehicle actuated (VA) system and a real-time reinforcement learning model (RLM) for non-lane following heterogeneous traffic.