LiDAR Spoofing Attack Detection in Autonomous Vehicles
Khattab M. Ali Alheeti, Abdulkareem Alzahrani, Duaa Al Dosary
Abstract
Detecting 3D objects is a core task for autonomous vehicles (AVs), as it allows them to drive safely and responsibly. To detect objects quickly and accurately, AVs have LiDAR sensors, which can capture 3D data from 360° in various conditions, including harsh weather. However, LiDAR sensors may face a spoofing attack via laser satirizing. This type of attack does not require physical contact with the sensors nor does it need to change them. It can mislead AVs by providing them with false information, which can then put passengers, pedestrians, and other vehicles in danger. Therefore, this paper proposes a model utilizing machine learning (i.e., decision trees) to protect AVs from such attacks. The performance of the proposed model indicates that it can detect LiDAR spoofing attacks.