Hybrid Control in Coordination of Multi-Agent Autonomous Vehicles at Intersections

Abstract

Over the years, traffic congestion has grown into one of today’s global problems. Intersections are a major cause of this problem. Thus proper management of the intersections will decrease this problem. In this study, an intelligent Intersection Control System (ICS) is proposed to control traffic flow in intersections, treating the problem as a Multi-Agent System (MAS). Through predicting the trajectories of the vehicles it is possible to minimize their travel times of vehicles while avoiding any predicted collision. In chapter (1), the problem of traffic congestion is introduced, as well as the emerging technologies of autonomous vehicles and the approaches of multi-agent systems, and concludes with problem statement of this study and the flow of the rest of the thesis. In chapter (2), a survey of the recent research efforts related to the focus of this study is presented; reaching to the conventional methods used to solve the problem under study. Relying on other researchers efforts, the approach to be used in this study is defined, as well as the contribution of this study. In chapter (3), the proposed methodology to tackle the problem is proposed through discussing the details of the different aspects of this system, and the new algorithm developed to quantify the collision situation in an intersection, and that is by using a hybrid controller composed of a both Fuzzy Logic controller and Genetic Algorithm. In chapter (4), the way the system was modeled is introduced, along with the simulation techniques used to verify the validity of the proposed approach, via a group of simulation experiments, and getting the results of these simulations. Chapter (5) presents a detailed description of the experimental platform developed to verify the algorithm proposed in chapter (3), and the details of its different modules. In chapter (6) the different set of experiments conducted on the experimental platform are described. The results achieved in these experiments are presented and discussed. Chapter (7) presents a conclusion to this study and its contribution, along with further research efforts that can be conducted. In this thesis a Hybrid Fuzzy-Genetic controller is developed to control intersections. Using the optimized hybrid controller, the intersection performance under different traffic capacities is studied. The results are compared with the existing traffic-light system. The results achieved reflect the improvement in the intersection utilization, where the intersection throughput increased by 91%, while the vehicles’ average and maximum delay times are decreased by 62% and 72% respectively.