Adversarial Attacks on Autonomous Vehicles: A Comprehensive Review

Growing advancements in artificial intelligence and autonomous vehicles have led us to a broader problem than just the functioning of these technologies. While autonomous vehicles (AV) have greatly improved in design and build to the point that we now have selfdriven taxis in some parts of the world, it doesn’t change the fact that there's a need for constant improvement in the system. This paper talks about the adversarial defense mechanisms in deep learning for autonomous vehicles, in particular, we will explore current research to gain a better understanding of progress and challenges within this field, as well as how adversarial attacks influence safety and critical choices made by autonomous vehicles. In addition to analyzing existing research, our review paper aims to offer novel insights into the long-term impacts of adversarial attacks on the functionality and reliability of AI models in autonomous vehicles. In essence, our review of 23 papers reveals a vital need to overcome the shortcomings of Autonomous Vehicles (AVs) and how there is a growing interest in studying adversarial attacks, which emphasizes the need of improving the safety and reliability of these systems.