Experimental Study Regarding an Intelligent Control System for Traffic Light Intersections

This paper aims to transform a simple traffic light intersection by technological improvement of the current equipment. Compared to a classical traffic light, the one made in this paper will also have a video camera connected to a minicomputer that will take images from traffic and will subject them to an analysis in order to identify the number of cars and the degree of traffic congestion. With this information, the decision may be made to modify the timing of passing or waiting color. The cars will be detected using a model previously trained by deep learning methods. Being able to make decisions based on traffic fluctuation, the new traffic light will be able to adapt to traffic conditions, thus managing to have an effect on the traffic congestion. The final results will highlight the experimental study and development of a low-cost application which will "over-equip" a traffic light with artificial intelligence in order to ease traffic.