Context-Driven Method for Smarter and Connected Traffic Lights Using Machine Learning with the Edge Servers

Abstract

In today’s world, traffic congestion has become a major issue in nearly all of the major cities around the globe. Due to an increase in the number of vehicles on the road, the intersections (or junctions) are major hotspots of congestion, so to maintain a proper flow, traffic intersections are installed with traffic lights that control the flow of the traffic moving in a particular direction. Traffic lights were first used nearly a century ago, but there have not been any major advancements in the functioning of traffic lights that can cope up with the requirements of today’s traffic. Nowadays, due to a lack of optimization in the timings of traffic lights, they can further cause more delays and congestions which causes environmental issues (like increased CO2 emissions and noise pollution), road rage, mental frustration, etc. This paper addresses the problem of traffic lights not having the optimized timings of signal for each direction due to the unavailability of robust, pro-active and real-time analysis of traffic and then analyzing and predicting the best timings of traffic lights based upon the size (small, medium, or large) and class (emergency and normal) of vehicles.