Enhancing Security of Autonomous Vehicles: Detecting Adversarial Attacks Targeting Traffic Signs

Autonomous vehicles (AVs) are revolutionizing the transportation sector; however, a huge setback to them becoming more prevalent in the market is the security vulnerabilities they are exposed to. The dependency of autonomous vehicles on machine learning and sensor-based systems makes them susceptible to many safety risks. These safety risks range from minor misinterpretations, such as failing to detect a trivial object, to critical failures like collisions or erratic driving. These vulnerabilities can undermine public trust and delay the widespread adoption of AV technology. This research explores a comprehensive approach to improve the robustness of AVs against adversarial attacks targeting traffic signs. This study investigates developing classification models to detect the proposed attack scenario. By addressing the root causes of erroneous actions of AVs, this research contributes to advancing secure and robust autonomous vehicle technologies essential for the safety of the public.