Smart Traffic Light System Design Based on Single Shot MultiBox Detector (SSD) and Anylogic Simulation

Traffic lights, which can optimize vehicle flow rates, solve congestion, and reduce accidents, are often found on every city road. At some intersections, the traffic light transition duration is still set manually without taking into account the number of vehicles, resulting in longer vehicle queues. This paper proposes a smart traffic light system that can overcome these problems. The system is divided into two main parts, namely detection of objects and determining the duration of the traffic light. In the object detection process, this system uses computer vision technology through the singleshoot multibox detector (SSD) algorithm to detect the number and average speed of passing vehicles. The data then become input for the Anylogic simulation to determine the optimal green light duration for traffic lights. After obtaining the optimal duration, this system will simulate the flow rate of the vehicles. In the trial stage, this system gave good results with the average number of vehicles increasing by around 156 vehicles with an average travel time of 6 s faster.