Traffic signal control using fuzzy logic and evolutionary algorithms

This paper presents a fuzzy control system to regulate the traffic flow approaching a single real intersection which consists of multiple lanes with turns, by adjusting time parameters and phases of traffic signals. The lanes are cataloged into several groups controlled by individual traffic lights. These lights are further arranged into several light phases. A fuzzy controller was developed to control the time length of each light phase. Evolutionary algorithms were employed to generate the fuzzy logic rule base, using real statistical traffic data for the intersection. To simulate real car flows, new acceleration and deceleration movement models were developed to ensure safe driving by avoiding possible collision. A new fitness function that comprehensively characterizes car flow delay induced from signals was constructed to evaluate the performance of the fuzzy logic controller. Key performance criteria obtained using the fuzzy logic controller were compared with those obtained by the controller used by the Department of Main Roads, Queensland at the intersection.