Autonomous vehicles are revolutionizing transportation, and the accuracy of road lane detection is a pivotal aspect of this innovation. This paper presents an in-depth exploration of a sophisticated lane detection system, geometric modeling to estimate the geometric structure of lane boundaries based on images captured by an onboard vehicle camera, and the deployment of object detection techniques. The lane detection system is meticulously designed, employing a series of computer vision techniques to identify and track lanes in various driving conditions. The curve fitting component utilizes a second-order polynomial, providing a mathematical model that accurately represents the curvature and intricate dynamics of the detected lanes. This mathematical representation provides a more nuanced understanding of the road geometry, aiding in the prediction of vehicle trajectory. The object detection facet of the research focuses on the recognition and classification of objects within the driving environment, contributing significantly to the overall situational awareness of autonomous driving systems. The YOLO (You Only Look Once) algorithm is commonly used for this purpose as it can process frames at an impressive speed while maintaining high accuracy, making it suitable for real-time applications. The efficacy of the suggested techniques was confirmed by conducting experiments on two distinct datasets. The proposed method achieved an accuracy of 98.64% on the Tusimple and 96.92% on the KITTI dataset, demonstrating its robustness and reliability under varying conditions.
 
Special Issue of AEIC 2024 (Electrical and System & Computer Engineering  Session)