Object Detection Techniques in Autonomous Driving

Abstract

This review comprehensively examines recent advancements in object detection (OD) methods for autonomous driving, highlighting their critical role in ensuring the safety and efficiency of autonomous vehicles in complex environments. It discusses various methodologies, including the application of machine learning (ML) techniques, and the integration of sensors like LiDAR and radar, which enhance the system’s ability to accurately identify and track nearby objects, such as pedestrians, vehicles, and obstacles, in real-time. The review synthesizes findings from multiple studies, showcasing innovations like adversarial learning techniques that improve detection performance, especially in adverse conditions. Furthermore, it addresses significant challenges, including environmental variability, computational efficiency, and the threat posed by adversarial attacks, which can compromise detection accuracy. The review emphasizes the importance of developing more robust and adaptive models, and it outlines future directions such as enhancing sensor fusion methods, optimizing model architectures, and employing open-world learning to prepare for unexpected scenarios, ultimately aiming to improve the reliability and safety of autonomous driving technologies.