ADAPTIVE INTELLIGENT TRAFFIC CONTROL SYSTEMS IN SMART CITY

Abstract

Traffic congestion in urban areas presents a significant challenge with far-reaching impacts on the economy, environment, and overall quality of life. To address this challenge, this thesis proposes a novel approach to traffic signal control aimed at alleviating traffic congestion more effectively. The research problem this study explores is the design and implementation of an adaptive system for traffic signal control in urban road networks, specifically focused on how to effectively manage traffic signal timings to mitigate congestion. The major contributions of this study include the development of a unique coordination algorithm for adaptive traffic signal control, utilizing Multi-Agent Reinforcement Learning (MARL) and Ant Colony Optimization (ACO). This algorithm's uniqueness is reflected in its capacity to simulate the behavior of ant colonies to guide multiple agents in managing traffic signals at various intersections, enabling them to learn from their environment and interactions to optimize signal timings By simulating the behavior of ant colonies, the algorithm guides multiple agents in managing traffic signals at various network intersections, learning from their interactions with the environment and each other to optimize signal timings. This research sets out to address the challenge of traffic congestion in urban areas. With cities worldwide struggling with this issue, the task of managing traffic signal timings to reduce congestion is paramount. The problem formulation involved the exploration of how novel Machine Learning (ML) techniques, such as Multi-Agent Reinforcement Learning (MARL) and Ant Colony Optimization (ACO), could be utilized to develop an adaptive coordination algorithm for traffic signal control. These techniques were chosen due to their potential for learning and adapting over time to optimize signal timings based on ever-changing traffic conditions. The novelty of this research lies in the unique combination of MARL, Actor-Critic (AC), and ACO techniques to develop an adaptive coordination algorithm for traffic signal control. By integrating these techniques, we've created a system where multiple agents can independently control traffic signals at different intersections, learning from their surroundings and interactions to continually improve signal timings. This innovative use of ML, especially MARL and ACO, represents a significant contribution to the field of traffic management, as it offers the potential to adapt to changing traffic patterns and conditions in real-time. This adaptability is expected to lead to more efficient traffic flow and decreased congestion, outcomes not fully realized by existing fixed-time and traditional adaptive signal control methods.