Roboticists from Northwestern University demonstrated an experimental unmanned aerial vehicle called Phantom Twist that utilizes rapid spinning to become nearly invisible to human eyes during flight, according to a research paper presented at Robotics Science and Systems 2026 in Sydney. Developed by a team including Jingxian Wang, Chen Yu, David Matthews, Emma Alexander, Sam Kriegman, and Michael Rubenstein, the micro-drone achieves an order of magnitude lower visibility than a standard quadrotor of comparable size by exploiting human persistence of vision and computationally optimized physical design.
“The design space is high dimensional,” Michael Rubenstein explained regarding the development process. “It’s very difficult for a human to reason through all the trade-offs between the physical constraints required for stable flight and the visual appearance of the spinning drone, and I don’t think we would have easily arrived at this low-visibility design ourselves.”
Phantom Twist operates by spinning rapidly in flight at speeds between 15 and 25 hertz. This motion transforms a solid physical object into an opaque smear. Human eyes typically require approximately 100 milliseconds to integrate visual scenes before sending processed imagery to the brain. When an object moves fast enough across a visual field, human eyes average the motion across the background, creating a transparent blur known as persistence of vision. Because the Northwestern drone is engineered with substantial empty space within its rotating footprint, the rapid rotation renders the craft exceptionally difficult for human observers to detect.
Computational Optimization and LPIPS Metrics
Constructed from 0.8-millimeter carbon fiber rods, batteries, a controller, counterweights, a motor, and a propeller, Phantom Twist features a flexible component arrangement managed by a handheld launcher. Controlling such a vehicle requires pulsing motor speed up or down during each individual rotation to dictate translational direction, while overall motor thrust controls altitude and passive stability stems from the spinning design.
To achieve minimal visibility, the research team employed an iterative optimizer designed to minimize a metric called learned perceptual image patch similarity, or LPIPS. LPIPS calculates the difference between a clean background image and that same background overlaid with a simulated spinning drone. Out of approximately 20,000 evaluated configurations, the finalized optimized design achieved an LPIPS score of 0.0104. By comparison, a human-designed configuration scored around 0.2, making it twice as visible, while a conventional quadrotor of identical size registers more than 10 times more visible.
“The automated pipeline prefers placements where components don’t visually overlap as it spins, or where the components are too close to the center of rotation,” Rubenstein noted regarding the resulting structural layout.
Future Applications and Outdoor Flight
Currently, Phantom Twist relies on an optical tracking system for stabilization and navigation, restricting its operation to indoor controlled environments. However, researchers have expressed optimism about transitioning the technology outdoors by drawing on prior lab work involving similar spinning flight principles. Beyond minimizing visual footprint, the rotating body opens possibilities for specialized payload integration.
“An interesting possibility is mounting a camera on the spinning body,” Rubenstein stated. “As the vehicle rotates, it could capture imagery in every direction, effectively creating a 360-degree view of its surroundings that could be used for onboard navigation and control.”
Potential future deployment scenarios span from covert surveillance applications to close-range wildlife monitoring. Researchers note that a quieter, visually imperceptible micro-drone could observe animal populations with minimal disruption to natural behaviors. Further details regarding the engineering methodology are available in the research paper titled Computational Design of a Low-Visibility UAV Using a Human-Aligned Perceptual Metric, detailed at the RSS 2026 conference.
Related reading