Australian researchers have unveiled a groundbreaking cyber algorithm designed to thwart man-in-the-middle (MitM) attacks on unmanned military robots, ensuring rapid shutdown in seconds
In a revolutionary experiment harnessing deep learning neural networks, experts from Charles Sturt University and the University of South Australia simulated human brain behaviour.
Cyber algorithm: Unprecedented success in real-time
They trained the robot’s operating system to recognise the hallmark of a MitM eavesdropping cyberattack, a tactic where assailants disrupt ongoing conversations or data transfers.
Deployed on a United States army combat ground vehicle replica, the algorithm demonstrated an astounding 99% success rate in foiling malicious intrusions. Meagre false favourable rates, less than 2%, underscored the system’s unwavering effectiveness.
The groundbreaking results of this research have been documented in IEEE Transactions on Dependable and Secure Computing, reaffirming its pioneering value.
A new standard in cybersecurity
Professor Anthony Finn of UniSA heralds the algorithm’s superiority over global cyberattack recognition techniques. Collaborating with the US Army Futures Command, he and Dr Fendy Santoso at Charles Sturt Artificial Intelligence and Cyber Futures Institute replicated a MitM cyberattack on a GVT-BOT ground vehicle, educating its operating system to detect and combat such assaults.
Robotic Operating Systems: A vulnerability
Prof. Finn emphasizes the heightened susceptibility of robotic operating systems (ROS) to data breaches and electronic hijacking, given their extensive networked nature. The rise of Industry 4, characterized by advancements in robotics, automation, and the Internet of Things, necessitates seamless collaboration among robots, exposing them to cyberattacks. However, he underlines the bright side: rapid advancements in computing power have made it feasible to develop and implement cutting-edge AI algorithms for a robust defence.
A secure future for robotics
Dr. Santoso points out that despite their widespread usage and numerous advantages, robotic operating systems largely overlook security concerns in their coding schemes. This stems from encrypted network traffic data and limited integrity-checking capabilities. Thanks to deep learning, their intrusion detection framework boasts remarkable resilience and accuracy, is adept at handling large datasets, and is perfect for securing data-driven systems like ROS.
Charting new frontiers
Prof. Finn and Dr. Santoso’s ambitions extend beyond ground vehicles. They plan to challenge their intrusion detection algorithm on various robotic platforms, such as drones, known for their swifter and more intricate dynamics than ground robots.
