Industrial environments are increasingly connected to the internet. This connectivity offers many advantages such as opportunities in predictive maintenance and remote monitoring, but also brings risks. For instance, robots can be attacked by cybercriminals. This can not only jeopardise production but, in the worst case, cause serious injury. Researchers at the Georgia Institute of Technology have built a device that detects and stops such attacks early.
This is HoneyBot, a robot named after the software-based 'Honeypots' from the IT security world. A honeypot poses as a computer system that is vulnerable to cyber attacks and viruses, but in reality has been groomed by security experts to gather information about cyber attacks. Among other things, honeypots help detect new attack campaigns and identify misuse of specific security vulnerabilities. HoneyBot is a robot that uses this approach to detect cyber attacks on robots in industrial environments and alert security experts about them so they can take immediate action.
Fatalities
"Robots are doing more than ever and some companies are already working on not just assembly robots, but free-standing robots that can drive across the production floor," says Raheem Beyah, a professor at Georgia Tech's School of Electrical and Computer Engineering. "In such an environment, you can imagine how dangerous it could be if a hacker managed to gain access to these machines. At the very least, they could damage the production being produced. However, if the robot is large enough, it could also destroy parts or the production line. In the worst case, it could cause injury to people in the area or even kill them."
HoneyBot is a small robot with wheels that fits into a shoebox. The robot presents itself in the network of a manufacturing environment as an industrial robot that plays a crucial role in a production, which makes it an interesting target for cybercriminals. If an attacker breaks into the HoneyBot, the robot collects information about the attack and the attacker. "Many cyber attacks go unpunished as rogue parties enjoy a certain degree of anonymity on the internet. This makes it difficult for companies to determine who is responsible," said Celine Irvene, a PhD student who worked with Beyah on HoneyBot. "Honeypots allow security professionals to study the attackers and determine what methods they use, where they are located and possibly even who they are."
Fake sensor data
HoneyBot feeds attackers fake sensor data to disguise its true identity and do its job. "If an attacker is smart and alert to honeypots, he might look at sensors on the robot such as an accelerometer or speedometer to verify that the robot is carrying out its commands. We therefore 'spoof' this information. This allows the hacker to see on the sensors that acceleration has taken place from point A to point B."
Researchers put the HoneyBot to the test in experiments, examining how convincing fake sensor data is to individuals controlling the robot remotely. Volunteers controlled the robot through a virtual interface during the experiments, without actually being able to see the robot. The HoneyBot was placed in a maze, after which participants were instructed to drive the robot out of this maze.
Cut-off routes
To encourage volunteers to break the rules, forbidden cut-off routes were present at certain points in the virtual representation of this maze that allowed faster escape from the maze. However, these cut-off routes were not present in the real maze. If a participant used a cut-off route, the robot came to a halt in the real maze. The volunteer was then presented with false sensor data indicating that the robot had taken the cut-off route and was continuing its route.
Post-experiment surveys show that participants who actually controlled the robot throughout the entire experiment and those who were presented with fake sensor data rated the data as equally reliable. "This is a good sign as it indicates that we are on the right track," concludes Irvene.
Author: Wouter Hoeffnagel
Source: Georgia Institute of Technology (pictured)
