A Dutch lab flew a rudderless drone by giving it pigeon reflexes

Most aircraft carry vertical tails or rudders to tame Dutch roll — the skater-like mix of yaw, sideslip and roll. The tail adds weight and drag, which costs airliners fuel and inflates radar signature, a problem for stealth aircraft. The rudderless B-2 bomber makes do with drag flaps at its wingtips, which aerospace engineer and biologist David Lentink of the University of Groningen calls inefficient: the efficient way is solving stability by generating lift, and that is what birds do.

In 1929 the German scientist Franz Groebbels argued birds fly like 'automatic airplanes', stabilizing through reflexes — tail flipping when the body pitches, wings extending asymmetrically when it rolls. Lentink's team tested the hypothesis by building backward from it. The first robot, Tailbot — fixed wings, a five-degree-of-freedom tail — held stable in a simulated turbulent wind tunnel under a reflex controller, but crashed outdoors. The fix was PigeonBot II: pigeon-like morphing wings that tuck and extend independently, nine servos, roughly the weight of a real pigeon, and real feathers.

A stock Pixracer drone autopilot only understands conventional quadcopter controls, so an Arduino sits between autopilot and robot, translating commands into wing and tail motions. Outdoors, PigeonBot II took off, flew and landed fully autonomously — or on high-level operator commands like up, down, turn — with flight stabilization resting entirely on bird-like reflexes. Groebbels' 1929 conjecture now has a flying demonstration.

One component still defeats engineering: the feathers themselves, with 10-micron 3D hooks forming the world's only one-sided Velcro, each locked hook bearing loads up to 20 grams. No manufactured structure is simultaneously that lightweight, stiff and complex, so the robots fly on real pigeon feathers.

Replace the rudder with the bird's answer: a reflex-like controller folds and spreads morphing wings and tail, stabilizing with lift instead of drag.

It attacks the tail's cost at the source — stabilizing with lift-generating shape change as birds do, rather than patching with wasteful drag flaps like the B-2.

A 95-year-old biology hypothesis became a flying engineering test: build the reflex loop and see if the theory stays airborne.

The avionics stayed commodity hardware: a shim Arduino let a stock drone autopilot command morphing wings and a five-degree-of-freedom tail.

Removing the rudder removes its weight, drag and radar signature in one stroke — the entire point of flying rudderless.

When a control surface costs too much, borrow the animal's solution: distributed shape change driven by fast reflexes can replace the part entirely.

PigeonBot II passed the outdoor test: autonomous takeoff, flight and landing, stabilized purely by the reflex controller, with an operator issuing only high-level commands. The team still flies real feathers because nothing engineered matches them; the result gives aircraft designers a lift-based route to rudderless flight, where a vertical tail is exactly the liability.