“Phantom Twist”: US Researchers Build a Drone That Blurs Before Your Eyes
Set Trending Topics as a preferred source on Google.
A team at Northwestern University has built a prototype that doesn’t camouflage itself but exploits the sluggishness of human vision instead: the entire aircraft rotates up to 25 times per second, turning it into a faint smudge.
Attempts to make drones invisible have so far relied mostly on camouflage patterns, transparent materials or light-bending optics. A team at Northwestern University’s McCormick School of Engineering in Evanston, Illinois, has taken a different route: rather than changing what the machine looks like, they designed around how it is perceived. The prototype, called “Phantom Twist,” uses motion blur — the same effect that makes fast-spinning fans and propellers appear to vanish.
The work was presented on July 16 at Robotics: Science and Systems (RSS) 2026 in Sydney, under the title “Computational Design of a Low-Visibility UAV Using Human-Aligned Perceptual Metric.”
One motor, one propeller — and a spinning body
Unlike a conventional quadcopter with four rotors, the drone runs on a single motor and a single propeller. The propeller turns one way, the rest of the airframe turns the other. The result: there are no stationary parts left for the eye to lock onto.
<cite index=”20-1″>”For our drone, the whole thing is rotating, so there are no stationary parts,”</cite> says project lead Michael Rubenstein, an associate professor of computer science and mechanical engineering. On a standard quadrotor, he notes, the propellers spin but the body stays visible.
Co-author Emma Alexander, a computer vision researcher, explains the effect through the way the eye works: it accumulates signals over time, roughly comparable to a camera’s exposure time. Spin an object fast enough and it loses its distinct features for the observer. Because the prototype is mostly open structure to begin with, its few opaque components are visually averaged with the background into a slight haze.
20,000 configurations, worked through algorithmically
The design process was largely automated. A computational model first generated some 20,000 configurations capable of stable flight. AI and optimization algorithms then repeatedly rearranged the main components — motor, propeller, circuit board, counterweight and batteries — to minimize visibility from virtually every viewing angle without sacrificing flight stability.
Promising candidates were simulated in flight and overlaid onto a hundred real-world backgrounds. A perception model approximating human vision assigned each design a visibility score; the 500 least conspicuous went into a further round of optimization. Only once a design met every criterion was it actually built.
The resulting layout distributes components at different heights, at different angles and with plenty of space between them, so they don’t visually overlap while spinning. Measured against the team’s own visibility metric, the prototype is roughly ten times less perceptible than a conventional quadcopter.
Still audible
That does not make the drone invisible. The researchers list several limitations themselves: the propeller is audibly loud, and wires and support rods remain partly visible. Future versions are meant to use more transparent materials and quieter propulsion.
The team frames the use cases as civilian — scenarios where drones alter the very thing they are observing: wildlife monitoring, environmental surveys and infrastructure inspection. Animals scatter and people behave differently once they notice an aircraft; a harder-to-see machine could do the same job with less disruption.
The work is best read as basic research: the optimization targeted visual perceptibility by humans and many animals only, not radar, infrared or acoustic signatures. The study was funded by the US National Science Foundation.
