Unmanned vehicle can be deceived by special pictures
The discovery poses a real security risk when using robotic cars on public roads.
Specialists from the Institute of Intelligent Systems. Max Planck and the University of Tübingen conducted a study of unmanned vehicles for safety. Engineers checked how cars cope with the recognition of human figures.
A total system failure occurs, an unmanned vehicle can leave the lane or brake unexpectedly. In some ways, this resembles the effect when stroboscopic flashes of a certain frequency cause epileptic seizures in some people. That is, through vision, a malfunction of the body's functions is artificially caused, in this sense, the human brain and the drone have something in common.
According to scientists, in just four hours they were able to create a sample of color combinations that cause an unmanned vehicle to a state akin to panic, and this becomes a security risk. The pattern can be easily applied to t-shirts or made stickers on road signs or shopping bags. Hackers can also take advantage of this, researchers warn.
The problem is the imperfection of artificial intelligence, which manifests itself in image recognition. The algorithm uses an integrated camera to observe the environment, such as the road in front of the car, and to detect obstacles. If recognition fails, the robot machine at best stops for security reasons.
The authors of the study emphasized that such a bug occurs with a probability of only a few percent, but this is enough to make the drone behave unpredictably. The experiment showed that if the car’s camera sees the same spot several times, its reaction will be special every time.
Of course, scientists and programmers will eventually solve this problem, but so far it remains. Researchers believe that now the challenge for automakers is to train their systems to be resistant to such attacks.