A Camera-Based Intelligent Traffic Signal Control System for Linear Cities: A Case Study of Hulhumalé, Maldives

Abstract

Traffic congestion is a growing challenge in Hulhumalé as it develops into a smart city. A major factor contributing to delays is the time vehicles spend idling at traffic lights, particularly during peak hours on the Sinamalé Bridge, where waiting times can reach up to two minutes per signal. Existing smart traffic solutions in other cities often rely on embedded sensors, which are expensive to install and disruptive to road infrastructure. To address these limitations, this project proposes a vision-based traffic light system capable of detecting vehicles in real time and dynamically adjusting green-light durations. By adapting to actual traffic conditions instead of relying on fixed schedules, the system aims to reduce waiting times and improve overall traffic flow. The concept will be tested using a prototype of a three-way junction simulated in Python with a custom trained YOLO-v8 for vehicle detection. Development follows a prototype-based experimental engineering methodology. The expected outcome is a scalable and cost-effective solution capable of reducing average waiting times by at least 30% compared to the current system, offering a practical step toward smarter and more efficient urban mobility