"We attack products so their manufacturers can fix problems and protect consumers before somebody malicious takes advantage," says Yan.
In this case, Yan's team was devising attack strategies where an adversary could mess with a vulnerability in drone-camera systems to wrest control of a drone from its pilot. Here's how they did it...
The researchers discovered that shining lights into a drone's camera prompted the drone software to kind of hallucinate and "see" the lights as an obstacle. As the drone's autonomous controls designed to avoid collisions engaged, the team could adjust the lights to control where the "phantom object" appeared and take over the steering of the UAV.
“We could manipulate the drone by controlling the angles and intensity of the light,” says Yan. “By controlling the location of the drone's 'hallucination,' we control which way the drone moved. Imagine that an Amazon delivery drone is under such an attack. Your packages would be effectively seized by the attacker, while the drone pilot has no idea why it’s happening."
The team successfully targeted quadcopters from enough distance away so that the attacks could avoid being detected by the true pilot. The action proved that companies using drones as delivery and inspection services could lose control of a quadcopter to thieves and forfeit property and data.
The good news is that the attack—which the MSU team named DoubleStar—can be thwarted with countermeasures such as blocking some of the light with lens hoods and updating the drone’s obstacle avoidance software.
“Once they know the existence of an attack, they can specifically tune their algorithms for defense,” says Yan. “We don’t think it will be very hard to do.”
In the meantime, if your drone starts staring into projector lights, you may have a far-out, tripping stoner on your hands.
