Thesis Defense Announcement for Mahfuzur Rahman – 10/18/2023 at 2:00 PM
October 10, 2023
Thesis Title: Traffic Light Detection and V2I Communications of an Autonomous Vehicle with the Traffic Light for an Effective Intersection Navigation using MAVS Simulation
When: 10/18/2023, 2:00 PM
Where: Simrall 228
Candidate: Mahfuzur Rahman
Degree: Master of Science in Electrical and Electronics Engineering
Committee Members: Dr. John E. Ball, Dr. Ali C. Gurbuz, Dr. Umar Iqbal
Abstract
Intersection Navigation plays a significant role in autonomous vehicle operation. For effective and accurate intersection navigation, an autonomous car needs to identify the traffic lights accurately, and at the same time, it needs to understand the time and phase of the traffic lights. This paper presents a comprehensive research initiative focused on enhancing the intersection navigation capabilities of autonomous vehicles through the integration of advanced computer vision and Vehicle-to-Infrastructure (V2I) communication systems simulating the scenarios in a high-fidelity automotive simulator. The research unfolds in two distinct but interconnected parts. In the first phase, we propose an approach utilizing YOLOv8, the latest iteration of the YOLO deep learning model, to meticulously detect and recognize the status of traffic lights. This process commences with the utilization of real-time local images to develop and test the YOLO network. Subsequently, the model undergoes intensive training on the SJTU Small Traffic Light Dataset (S2TLD) to ensure robust and precise recognition of traffic light states. The second phase focuses on V2I communication, establishing seamless connectivity between autonomous vehicles and traffic lights. Within a simulated Mississippi State University Autonomous Vehicle Simulation (MAVS) environment resembling a small city with multiple intersections, we create a traffic light control system enabling the transmission of Signal Phase and Timing (SPaT) messages to vehicles. These messages encompass traffic light phases and timing details for phase changes, allowing autonomous vehicles to autonomously adjust their speed and behavior in real time. The simulation result demonstrates that the autonomous car detects the traffic light accurately and at a certain distance closer to the nearest intersection, it receives a short message indicating the current phase of the light and the time to change into the next phase. The research extends to simulating multiple intersections, simultaneously observing the interactions between the vehicle and traffic light infrastructure.