Researchers at security firm IOActive have shown how a remote attacker can hack an industrial collaborative robot, or cobot, and modify its safety settings, which could result in physical harm to nearby human operators.
A few months ago, IOActive published a brief report providing a high-level description of its research into robot cybersecurity. Researchers analyzed industrial and business robots from six vendors, including SoftBank Robotics, UBTECH Robotics, ROBOTIS, Universal Robots, Rethink Robotics and Asratec Corp.
A brief analysis of mobile applications, software and firmware led to the discovery of nearly 50 vulnerabilities, including weaknesses related to communications, authentication, authorization mechanisms, cryptography, privacy, default configurations, and open source components.
Cesar Cerrudo, CTO at IOActive, and Lucas Apa, Senior Security Consultant at IOActive, warned at the time that the security holes could be exploited for spying and stealing sensitive data, and even cause physical damage or harm in the case of industrial robots. They have now shared technical details and videos demonstrating some of their findings.
Cobots share a workspace with human operators and help them perform various tasks. Unlike traditional industrial robots, which execute repetitive tasks, they can learn new movements, see using cameras and hear via microphones.
Cobots are used not only in industrial environments, but IOActive’s research has focused on industrial models, namely the Baxter/Sawyer cobots for industrial automation from Rethink Robotics and UR from Universal Robots.
Cerrudo and Apa showed how a remote attacker could chain six vulnerabilities to modify a UR robot’s safety settings and disable emergency functions, which could pose a serious threat to human lives.
“Imagine what could happen if an attack targeted an array of 64 cobots as is found in a Chinese industrial corporation,” the researchers warned.
Technical details have been provided for each of the exploited flaws, along with a video showing how an attacker can disable safety features and cause a robot arm to act “crazy.”
While some of the robot manufacturers contacted by the IOActive researchers have taken steps to address the vulnerabilities, others have downplayed the risks and did not release any patches. In the case of cobots, Rethink Robotics fixed the flaws discovered by the experts back in February, but UR still hasn’t resolved the issues affecting its products.
IOActive is not the only security firm to analyze industrial robots. Researchers at Trend Micro and the Polytechnic University of Milan published a paper a few months ago on the cybersecurity risks associated with industrial robots, and warned that some machines could be attacked directly from the Internet.
Elon Musk, CEO of Tesla and SpaceX, along with more than 100 robotics and artificial intelligence entrepreneurs recently sent a letter to the United Nations calling for action to prevent the development of robotic weapons.
Related: Vulnerabilities Found in Double Telepresence Robots