Development and evaluation of a multi-agent autonomous vehicles intersection control system

The problem of traffic congestion is one of today´s major problems all over the world. Traffic congestion increases fuel consumption, pollution, and decreases productivity. Intersections congestion is a main cause of such problem. Intelligent Transportation System (ITS) approaches try to solve these problems by integrating artificial intelligence in the traffic networks. This study proposes an Intersection Control System (ICS) based on a Multi-Agent System (MAS) structure, to control Autonomous Vehicles (AVs) in the intersections for better roads utilization. A central Intersection Manager Agent (IMA) is implemented at each intersection while each vehicle is controlled by a Driver Agent (DA). Through the predicted trajectory idea, the ICS can increase the traffic capacity of the intersection while avoiding any predicted collision. The proposed approach is verified using a developed simulation environment. Both PID and Fuzzy controllers are tested. The results achieved are promising, the intersection throughput is increased by 64%, while the average and maximum delay times decreased by 51% and 62% respectively.