In a laboratory, the researchers were able to change the configuration of an ABB IRB140 robot, introducing an error that caused the machine to change how it operated. Although the effect – shown in the video below – was minor (moving a line drawn by the arm by 2mm), the researchers point out that this could be enough for a product to fail or behave unexpectedly.
The researchers found several weaknesses in industrial robot architectures and implementations, and identified five types of attack that could exploit these flaws. They include:
• altering the control system so that the robot moves unexpectedly or inaccurately;
• changing the robot’s calibration to make it move unexpectedly or inaccurately;
• manipulating a program executed by the robot to introduce defects in a workpiece;
• manipulating status information so that the operator is not aware of the true status of a robot; and
• manipulating the status of a robot so that the operator loses control or can get injured.
The researchers found that robot software is often outdated or based on vulnerable operating systems and libraries (and sometimes relying on obsolete or cryptographic libraries), and that robots have weak authentication systems with unchangeable default credentials.
The researchers suggest some ways that attackers could exploit these weaknesses. They could, for example:
• sabotage products by introducing minor defects that would cause the product to malfunction;
• alter products and then contact the manufacturer asking for ransom to reveal which lots were affected;
• disable or alter safety devices, causing a robot to damage itself or injure people;
• cause a robot arm to behave erratically, damaging a production line or causing production bottlenecks; or
• extract sensitive data – such as source code, or information about production schedules.
In another part of their project, the Trend Micro researchers found that tens of thousands industrial devices, including robots, reside on public IP addresses, which attackers could access and compromise.
The analysts suggest that robot standards need to consider cyber-security threats in a similar way to standards for industrial controls and the automotive sector. They point out that robots have long lifetimes and that their vendors should provide security updates to all currently deployed versions – which they may not always be able to do. In addition, users worried by downtime caused by updates may not patch their systems often enough.
As part of their project, the Trend Micro and Polimi researchers contacted robot-makers and they report that the manufacturers “have taken our results very responsibly, showing a positive attitude toward securing the current and future generation of industrial robots”. In particular, ABB Robotics stood out by welcoming their suggestions and starting work on a response plan for its current products.
Trend Micro and Polimi have published a free 47-page report on their research, which offers suggestions on how to enhance robot security.