Machine vision algorithms applied to dynamic traffic light control

Abstract

This paper presents a fuzzy traffic controller that in an autonomous, centralized and efficient way, manages vehicular traffic flow in a group of intersections. The system uses a computer vision algorithm to detect the number of cars in images captured by a set of strategically placed cameras at every intersection. Using this information, the system selects the sequence of actions that optimize traffic flow within the control area, in a simulated scenario. The results obtained show that the system reduces the delay times for each vehicle by 20% and that the controller is able to adapt smoothly to different flow changes.