Australian researchers have created a cutting-edge algorithm capable of swiftly countering man-in-the-middle (MitM) cyberattacks on unmanned military robots, effectively shutting them down in seconds.
Using deep learning neural networks to mimic human brain behavior, artificial intelligence experts from Charles Sturt University and the University of South Australia (UniSA) have trained a robot’s operating system to recognize the signature of a MitM eavesdropping cyberattack, where attackers disrupt ongoing conversations or data transfers.
In real-time experiments conducted on a replica of a US Army combat ground vehicle, the algorithm displayed a remarkable 99% success rate in preventing malicious attacks. The system’s false-positive rate was under 2%, validating its effectiveness. Their findings have been published in IEEE Transactions on Dependable and Secure Computing.
Improved Cybersecurity
Professor Anthony Finn, a UniSA autonomous systems researcher, explained that the proposed algorithm surpasses other global recognition techniques in detecting cyberattacks. He and Dr. Fendy Santoso from the Charles Sturt Artificial Intelligence and Cyber Futures Institute collaborated with the US Army Futures Command to replicate a MitM cyberattack on a GVT-BOT ground vehicle and trained its operating system to identify an attack.
The Vulnerability of Robot Operating Systems
Professor Finn emphasized that the robot operating system (ROS) is highly susceptible to data breaches and electronic hijacking because of its extensive networking capabilities. Industry 4’s emergence, marked by advancements in robotics, automation, and the Internet of Things, necessitates robots to work collaboratively, rendering them highly vulnerable to cyberattacks.
The Role of Deep Learning
However, the researchers revealed that the acceleration of computing capabilities allows for the development and implementation of advanced AI algorithms to safeguard systems against digital attacks. Dr. Santoso explained that despite its benefits and widespread use, the robot operating system frequently overlooks security concerns in its coding scheme due to encrypted network traffic data and limited integrity-checking capabilities.
Their intrusion detection framework, powered by deep learning, is robust and highly accurate, capable of handling large datasets suitable for protecting large-scale real-time data-driven systems, like ROS.
Future Testing
The researchers are planning to evaluate their intrusion detection algorithm on different robotic platforms, including drones, with faster and more complex dynamics compared to ground robots.
As we celebrate Cyber Security Awareness Month in October, this development marks a significant advancement in ensuring the security of robotic systems, especially in military and other critical applications.
