Intelligent Attack Detection in ROS-based Systems

摘要

Robotic systems, which were initially intended only for industrial work, have become a part of our daily lives in recent years. ROS is a popular middleware for developing applications in robotic systems. The security of ROS-based systems has gained importance with the expansion of robotic system usage. To enhance security, machine learning algorithms play an important role in robotic systems, especially in attack detection. This article provides an overview of the security of robotics by examining the machine learning based detection methods of cyber-attacks against ROS. We start by presenting security issues with a brief review of attack vectors in ROS. Next, we discuss the machine learning based attack detection approaches. Finally, we conclude by reviewing studies in the literature that use machine learning algorithms on attack detection in ROS. This survey aims to investigate the most popular and effective techniques applied to detect cyber-attacks, focusing on highlighting the latest advances in the field of machine learning in ROS attack detection. Our study will raise awareness among researchers in robotic system security and encourage researchers to upgrade attack detection in ROS-based systems.