NEAT-Traffic: An Evolutionary Network based approach towards Intersection Optimisation

In densely populated areas and metropolitan cities like Delhi, Mumbai etc. there are many vehicles on road for purpose of daily commute, Goods transportation, essential services delivery etc. Most of these vehicles have different set of starting, intermediate and final destinations and to regulate all this traffic, traffic intersections and traffic lights play a vital role. Most of the things have started becoming AI powered, real-time computation based these days, still the Traffic light timings are based on historical surveys and are not adaptive to sudden changes in traffic flow and just change to pre-fed timing in the system at specific time of the day. It can be observed on regular basis that at some intersections, one road remains heavily populated with vehicles and enough green light time is not allotted to it for allowing major portion of traffic to pass whereas some other road might have very less traffic yet comparatively much more green light time is allotted to it than what is actually needed. Moreover, when traffic volume is extremely low like in early morning or late night, traffic vehicles need to wait on Traffic Intersection for their turn even if no one is present on other roads of the intersection.