- Prototype drone spins its entire body up to 25 times per second to blur itself into near-invisibility
- Phantom Twist's single motor spins the whole airframe in one direction and its propeller in the other
- AI design pipeline tested 20,000 possible layouts to find the optimal configuration
Drones have a visibility problem. Whether they're filming guests at a wedding or hunting for candid shots of nesting birds, the moment a drone is noticed, it can change the behavior of whatever it's observing — wildlife scatters, people act differently, and you end up with iffy results.
A team of engineers at Illinois' Northwestern University may have found a way around that, and it has nothing to do with camouflage. Their prototype drone, dubbed "Phantom Twist," doesn't try to look like its surroundings. Instead, it spins itself into a blur so fast that the human eye simply can't keep up.
The drone rotates up to 25 times per second — well beyond what the eye can resolve into a distinct shape. The drone doesn't disappear outright, but becomes what the researchers describe as a faint, semi-transparent smudge that blends into the sky or landscape behind it. It's not invisibility in the sci-fi sense, but according to the team it's about ten times harder to spot than a conventional quadcopter.
You spin me right round
The trick lies in the design itself. A typical quadcopter has four rotors spinning while its main body stays completely still, which is exactly why you can still pick it out against the sky — the frame doesn't move, so your eye locks onto it. Phantom Twist ditches that setup entirely. It runs on a single motor and single propeller, with the propeller spinning one way and the rest of the drone's body spinning the opposite way. There's no stationary part left for your eye to latch onto.
Building something that flies stably while spinning that fast isn't a simple engineering job, though, so the team leaned on AI to do the heavy lifting. The process started with around 20,000 computer-generated drone layouts, each one checked for steady flight. From there, optimization algorithms repeatedly shuffled the position of the drone's key components — motor, propeller, circuit board, battery and counterweights — so they'd never overlap visually as the whole thing spun.
Every candidate design was then simulated spinning against 100 real-world photo backgrounds and scored using a model built to approximate human vision, with the 500 lowest-scoring designs pushed through further rounds of optimization before the final version was built.
The result spreads its components out at different heights and angles, so when it's in motion, everything blurs into a soft haze. It's clever stuff, and a reminder of how much of what we "see" comes down to how our eyes process motion over time rather than snapping a perfect still image.
It's worth being clear about what Phantom Twist is and isn't, though. This is a university research prototype, not a product with a release date or price tag, and it's still some way from ready for real-world deployment.
The propeller is audible even if the body isn't very visible. There's also no camera or imaging payload on the current build, which raises an obvious question for its stated wildlife-monitoring and infrastructure-inspection ambitions. It feels like a drone spinning 25 times a second would need some serious image stabilization or de-rotation trickery before it could shoot anything usable. The team hasn't detailed how (or whether) they plan to solve that yet.
There's an inevitable privacy angle to any drone that's harder to see, too, and it's not hard to imagine less charitable uses for the same principle down the line. For now, though, this remains a lab experiment funded by the National Science Foundation, aimed squarely at reducing wildlife disturbance rather than sneaking up on anyone — and there's no indication the spinning-body approach is heading toward consumer hardware anytime soon.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.
