Date of Defense
8-11-2024 10:00 AM
Location
E1-1038
Document Type
Thesis Defense
Degree Name
Master of Science in Software Engineering
College
College of Information Technology
Department
Computer Science
First Advisor
Dr. Salah Bouktif
Keywords
Traffic Signal Control, Urban Traffic Light Scheduling Problem, Deep Reinforcement Learning, Metaheuristics, Intelligent Transport Systems, Traffic lights, Traffic Signal Optimization.
Abstract
Managing road traffic in metropolitan cities is a crucial aspect of Intelligent Transportation Systems (ITS). The rapid growth of population and vehicles has led to increasing traffic congestion, which negatively affects travel times, fuel consumption, and air quality in urban areas. Intersections and their traffic lights are key contributors to this congestion, making efficient and adaptable Traffic Signal Control (TSC) and Traffic Signal Scheduling (TSS) essential. TSC manages traffic flow at intersections, while TSS optimizes the timing and sequencing of traffic signals. The techniques, Reinforcement Learning (RL) and Metaheuristic Optimization (MO), have shown promising results in addressing traffic control challenges individually. RL offers adaptability and learning capabilities, while Metaheuristics provide optimization through various problem-solving algorithms. By leveraging the complementary strengths of these approaches, the research seeks to create a more efficient, adaptable traffic control system. The study begins with a thorough review of existing TSC and TSS methodologies, followed by a comparative analysis of RL and Metaheuristic approaches specifically applied to traffic management. This comparison highlights the strengths and weaknesses of each method, offering insights into their performance in real-world scenarios. The research also delves into previous work on combining RL with Metaheuristics, exploring the benefits of such hybrid approaches. Furthermore, the study involves the development and implementation of a novel system that integrates both approaches to improve traffic signal control at congested intersections. The performance of this hybrid system will be validated through traffic simulations, with results compared against traditional methods and individual implementations of RL and Metaheuristics. The goal is to determine whether the combined approach provides measurable improvements in real-world scenarios or if standalone techniques are more effective, ultimately aiming to reduce congestion and improve urban transportation efficiency.
Included in
HYBRIDIZING REINFORCEMENT LEARNING WITH METAHEURISTICS FOR IMPROVED TRAFFIC SIGNAL CONTROL AND OPTIMIZATION IN URBAN TRANSPORTATION NETWORKS
E1-1038
Managing road traffic in metropolitan cities is a crucial aspect of Intelligent Transportation Systems (ITS). The rapid growth of population and vehicles has led to increasing traffic congestion, which negatively affects travel times, fuel consumption, and air quality in urban areas. Intersections and their traffic lights are key contributors to this congestion, making efficient and adaptable Traffic Signal Control (TSC) and Traffic Signal Scheduling (TSS) essential. TSC manages traffic flow at intersections, while TSS optimizes the timing and sequencing of traffic signals. The techniques, Reinforcement Learning (RL) and Metaheuristic Optimization (MO), have shown promising results in addressing traffic control challenges individually. RL offers adaptability and learning capabilities, while Metaheuristics provide optimization through various problem-solving algorithms. By leveraging the complementary strengths of these approaches, the research seeks to create a more efficient, adaptable traffic control system. The study begins with a thorough review of existing TSC and TSS methodologies, followed by a comparative analysis of RL and Metaheuristic approaches specifically applied to traffic management. This comparison highlights the strengths and weaknesses of each method, offering insights into their performance in real-world scenarios. The research also delves into previous work on combining RL with Metaheuristics, exploring the benefits of such hybrid approaches. Furthermore, the study involves the development and implementation of a novel system that integrates both approaches to improve traffic signal control at congested intersections. The performance of this hybrid system will be validated through traffic simulations, with results compared against traditional methods and individual implementations of RL and Metaheuristics. The goal is to determine whether the combined approach provides measurable improvements in real-world scenarios or if standalone techniques are more effective, ultimately aiming to reduce congestion and improve urban transportation efficiency.