This week, two research teams showed that the standard quadcopter design is far from the only way to build a flying robot. At Northwestern University, engineers unveiled a drone that spins its entire body at 1,500 RPM to become nearly invisible to the human eye. At MIT, a separate group demonstrated a flapping-wing robot that can swim underwater and launch itself into the air — just like a diving bird. Both projects challenge assumptions about what a flying machine should look like and how it should move.
The Phantom Twist: Designing for the Eye, Not for the Air
The drone from Northwestern's McCormick School of Engineering, called the Phantom Twist, doesn't try to hide by looking like a bird or a bug. It hides by exploiting a quirk of human vision called motion blur — the same effect that makes a fast-spinning fan or propeller seem to disappear.
Most stealth-drone work focuses on camouflage, transparent materials, or light-bending optics. The Northwestern team, led by associate professor Michael Rubenstein, took a different angle. They asked whether they could design the drone's physical layout around how people perceive motion.
"Invisibility has been pursued through materials and coatings," Rubenstein told IEEE Spectrum at the Robotics: Science and Systems conference in Sydney on July 16, where the team presented their paper. "We pursued it through motion."
The result is a single-motor, single-propeller aircraft that rotates 15 to 25 times per second. Its one propeller spins in one direction while the rest of the craft spins in the other. Because nothing stays still, there are no stationary outlines for the eye to track. The drone essentially blurs itself into a faint, semi-transparent haze against whatever background it's flying over.
To arrive at the final design, the team built a computational model that generated roughly 20,000 possible configurations capable of stable flight. An AI optimization algorithm then reranked those designs, simulated them spinning against 100 real-world backgrounds, and scored each one for visibility using a perception model that mimics the human visual system. The 500 lowest-scoring designs went through another round of optimization, where the algorithm repositioned components — motor, batteries, circuit board, counterweights — at different heights and angles to minimize visual overlap when everything is spinning.
The final prototype scored about 10 times less visually perceptible than a conventional quadcopter of similar size, according to the team's metric. On the other hand, it's still a lab prototype. The Phantom Twist relies on an external optical tracking system and has only flown in controlled indoor environments. Its wires, support rods, and optical-tracking tags remain visible. And the single propeller produces a noticeable whine — you can hear it coming even if you can't quite see it.
Still, the design process itself is a contribution. Rubenstein's group gave software the authority to place real hardware components, then built exactly what the algorithm said. The drone's motion blur is visible proof that automated design can produce functional machines whose geometry follows objectives that humans rarely optimize for directly.
Northwestern points to wildlife monitoring, environmental surveys, and infrastructure inspection as possible use cases — anything where a drone that blends in could observe without scaring off animals or drawing attention. The same low-visibility property also has obvious surveillance implications, a dual-use reality the team acknowledges but hasn't explored publicly.
The study was supported by the National Science Foundation.
The MIT Robot That Swims and Flies Like a Puffin
Across the country at MIT, a team led by assistant professor Raphael Zufferey took inspiration from the roughly 100 species of birds that can both fly through air and swim underwater. Loons, puffins, gulls, and petrels all do it naturally — they plunge into water to chase prey, then burst back into the air to fly away. Zufferey's group built a machine that copies the trick.
Published July 9 in the journal Science, the flapping-wing aerial-aquatic vehicle, or FAAV, weighs less than 300 grams — about half a pound. It has a central fuselage, two flexible flapping wings, and a steerable tail. The wings can be swapped for different sizes, and the team tested three sets: 60 centimeters wide, 80 centimeters, and 100 centimeters.
In experiments conducted in a water tank and later in Lake Geneva in Switzerland, the researchers identified the exact combination of wing size, flapping frequency, and tail pitch angle that lets the robot transition smoothly from swimming underwater to breaking the surface and flying. The magic number turned out to be medium-sized wings, a flapping frequency of around five beats per second, and a tail pitched at 70 degrees — steep enough to keep the wingtips from slapping the water on takeoff.
The robot swims underwater at about 1 meter per second and flies through air at about 6 meters per second. Both speeds and flapping frequencies closely match those of actual diving birds, which the team compiled from the scientific literature on puffins, kingfishers, and petrels.
One surprise: birds like puffins use their feet to paddle at the surface when taking off. The MIT robot doesn't need feet at all. "The question was, do we need the same for robots? And it turns out we don't," Zufferey said.
The FAAV's wings are made of thin membranes coated with hydrophobic nanoparticles that shed water on exit. The entire vehicle is waterproof and battery-powered, with a single electric motor driving a crankshaft that pumps the wings up and down.
Zufferey, who heads MIT's AURA Lab, envisions the robot being deployed for oceanography, marine biology, and coastal monitoring. "You could send this out not just every week, but every hour," he explained. "It could fly out at high speeds, dive in, fly back, deliver its data, and go back out, multiple times." The team sees applications in harmful algal bloom tracking, fish stock assessment, and infrastructure inspection under icebergs or at port facilities.
The work was supported by a Marie Skłodowska-Curie Actions fellowship grant.
The Bigger Picture: Robotics Is Leaving the Quadcopter Behind
The Phantom Twist and the FAAV share more than a publication date. Both represent a growing trend in robotics toward designs that are optimized for a specific capability rather than adapted from a general-purpose platform. The Northwestern drone sacrifices range, payload, and quiet operation for near-invisibility. The MIT robot trades speed in air or water for the ability to operate in both.
Both projects also relied heavily on computational design tools — optimization algorithms, perception models, and simulation — to explore design spaces too large for a human team to exhaust manually. Rubenstein's group iterated through 20,000 configurations; Zufferey's group tested three wing sizes, a dozen flapping frequencies, and multiple tail angles before finding the winning combination. That's a lot of data for a machine that weighs less than a bag of chips.
These advances come at a time when the robotics industry is expanding fast. The International Federation of Robotics reported that global industrial robot installations reached an all-time market value of US$16.7 billion in 2025, and new form factors — humanoids, micro-drones, amphibious vehicles — are driving the next wave of growth. The kind of specialized, computationally designed robots that Northwestern and MIT are building today could become the standard approach tomorrow.
Neither the Phantom Twist nor the FAAV is ready for commercial sale. The Phantom Twist needs an external tracking camera to stay aloft; the FAAV hasn't been tested in wind or waves. But the engineering logic behind both is already influencing how other labs think about robot design.
Sometimes the best way to make a drone disappear is to keep it spinning. Sometimes the best way to explore the ocean's edge is to build a robot that flies like a puffin. And sometimes the best way forward is to stop copying the quadcopter and start asking what a machine really needs to do.
Sources: IEEE Spectrum on Phantom Twist, MIT News on FAAV, Northwestern News, International Federation of Robotics
Read more about Robotics and Drones or the prior coverage of humanoid robots in factories.