FlexiQuad's flexible ring survives crashes and tight spaces

By Tom Whitfield

A 14-ounce FlexiQuad prototype developed at ETH Zürich withstood head-on impacts at about 11 mph without damage, and researchers report a fourfold advantage in collision resilience over a conventional quadcopter.

Almost every conventional quadcopter uses four stiff arms attached to a central body that holds batteries, control electronics, sensors and sometimes payloads. FlexiQuad removes that core: four rotors sit on a flexible ring, while batteries and other electronics are distributed evenly around its rim, leaving a hollow, vertically flattened center.

What does the ring change?

ETH Zürich researchers Luca Girardi and Stefano Mintchev, a former professor at ETH Zürich’s Environmental Robotics Lab, developed the aircraft. Girardi had pursued the concept for about five years and initially doubted it could fly without central reinforcement. Their model drew on bird and insect wings, which bend in some directions but resist deformation in others.

Engineers call that directional resistance anisotropic stiffness. “The key was to rethink where the weight sits, as well as where the structure bends,” Girardi said. “A heavy central body supported by flexible arms can introduce bending and oscillations.”

Moving batteries next to the motors eliminated the need for a central core, and Girardi described the hollow center as a consequence of that redistribution. Simply softening a conventional quadcopter’s arms would bend them under vertical rotor thrust, making the aircraft difficult to control.

FlexiQuad instead lets the ring flex sideways during a collision while remaining rigid vertically, so the propellers lift the body without wobble or sag. The structure can therefore absorb a sideways impact without compromising vertical thrust.

The material must also occupy a narrow range. Excess softness would let the propellers overlap during aggressive maneuvers; excessive stiffness would restore the breakage risk. Researchers found a range overlapping the stiffness of wings from tiny insects to large birds, and they say the approach scales from below 1 ounce to about 6.6 pounds.

What do the measurements show?

In a preprint posted to arXiv and not yet peer reviewed, Girardi and Mintchev report simulations in which an otherwise identical drone with one central battery produced 13 times the shaking at the rotors and eight times the shaking at the electronics during acceleration. The distributed-mass layout also spreads collision energy through the airframe instead of concentrating it centrally.

The 14-ounce prototype is roughly 1,000 times bendier than a conventional quadcopter. It tops 50 mph and accelerates at more than three times gravitational force. Girardi rejected the assumption that softness necessarily reduces performance: the aircraft can tip into a sharp sideways dodge and remain as maneuverable as a regular drone.

After the ring deforms and rebounds, the aircraft can stabilize and resume flight. In tests, FlexiQuad tolerated head-on hits at about 11 mph without sustaining damage. Demonstration footage shows it rebounding from a tree trunk before continuing along its route.

As the ring folds and the propellers overlap, thrust efficiency falls from 79% to 68%, although the aircraft keeps flying. FlexiQuad can squeeze through an opening 70% as wide as its normal width, and one test passed through a 7-inch slot.

The same deformation provides a grip. FlexiQuad hugged a soft first-aid pouch, carried it and released it; another test lifted a payload heavier than the aircraft. For transport, its ring opens flat inside a medium-size suitcase, and slide-in batteries plus quick-release fasteners simplify reassembly.

The inventors estimate low manufacturing cost. Four thin fiberglass-laminate strips, the low-cost material used in circuit boards, form the frame, with raw material cost below $15 before motors, electronics and batteries.

What remains before deployment?

Autonomy and imaging remain unresolved. The prototype still needs environmental perception and autonomous navigation, while post-impact body oscillation can blur camera images. The team has already used reinforcement learning and motion sensors to create a post-crash reflex that steadies the aircraft and resumes flight without pilot input.

The inventors also propose using deformation as a sensory channel, comparable to insects detecting changes in their wings. ETH Zürich nominated FlexiQuad for its Spark Award for invention of the year. Girardi and Mintchev have also applied for a patent covering flight where contact with nearby structures is difficult to avoid.

Proposed applications include:

• Flying through cramped ship ballast tanks
• Entering collapsed buildings during search-and-rescue missions
• Taking measurements in dense forests

The team is commercializing the design through a startup that intends to automate rapid drone design and coordinate closely with suppliers rather than sell only FlexiQuad. The prototype sits at NASA technology-readiness level 5 of nine, meaning researchers have validated it in realistic conditions, and the team is working toward level 6.

Girardi identified three commercial barriers:

• Converting the prototype into a commercial product
• Scaling manufacturing with a positive margin
• Ensuring customers will buy at the resulting price

A larger FlexiQuad with a net across its hollow center is about to enter a field pilot program, with the war in Ukraine cited as a test environment for catching kamikaze and rogue drones in civilian settings. Girardi said the approach suits locations “where you can’t afford kinetic interception,” because falling debris could strike people or sensitive infrastructure.

The team is also exploring gripping configurations with firefighters for search and rescue. If the field pilot and the move from readiness level 5 to 6 advance, the next evidence for FlexiQuad will be repeatable field performance, positive manufacturing margin and customer acceptance at the resulting price.