USE OF EVOLUTIONARY ALGORITHM TO OPTIMIZE ROAD TRAFFIC FLOW

Abstract

Due to increase in amount of vehicles traversing an intersection, it is difficult to handle them effectively. Although traffic lights are used to control the flow of these traffics in intersections, but these are not well enough as they need improvements in their algorithms to better control the traffic lights. This paper compares four different types of methods used in traffic control systems to reduce waiting times of these vehicles in intersections. Pretimed, deterministic, reinforcement learning and grey wolf methods are applied for uniform and varied demands. Test are done on a four-way intersection with multiples demands of traffic. The program of Simulation of Urban Mobility (SUMO) is used to test methods. Deterministic method performed best for lower demands while Grey wolf method performed best for higher demands considering uniform demands. While in case of varied demands, Grey wolf method performed best among all applied methods. Deterministic method performed 2nd best, pretimed method performed 3rd best and reinforcement learning method performed worst among all applied methods. The reasons behind these results is that the deterministic method possess knowledge about movement of vehicles which helps it to perform good for controlling flow of traffic which requires prior information of situations.