The Pilot Is the Fuselage: How a Deployable Rigid Wing Falls Out of Its Own Constraints

The gap between a wingsuit and a hang glider has no occupant

A wingsuit pilot in stable flight gets a glide ratio of roughly three to one. Three meters forward for every meter down, give or take, depending on the suit and the body in it. A rigid hang glider does far better, sixteen to one in good conditions, but the price of that number is infrastructure: a launch site, a tow or a cliff, a wing too large to wear through a doorway, and a recovery plan for wherever the flight ends. Between those two figures sits a wide band of performance that nothing currently occupies, because the things that fly well do not pack, and the things that pack do not fly well. A wing that could be carried in freefall posture, snapped open in under a second, flown at ten to one, and then set down under a reserve canopy would live in that band. The question is whether the physics allows such a thing to exist without killing the person wearing it, and if it does, what shape it is forced into.

MANTA is a development program built around that question. It is a wingsuit-extension rigid wing: a fitted harness with a carbon-fiber spine yoke, spars that run along the pilot's arms and legs and brace into position as the body spreads to a flight pose, and telescoping tip extensions that snap out from the wrists and ankles to complete a span the human frame cannot reach on its own. The target is a best-glide ratio near ten to one from a six-and-a-half square meter wing spanning six point three meters tip to tip, deployable from a freefall posture in about six tenths of a second. What makes the project worth writing about is not the target. It is the degree to which almost every major decision in the design turns out to be dictated by a constraint rather than selected from a menu. The wing is, in a real sense, solved for, not styled.

Treating the pilot as structure rather than payload

The first decision that looks like a choice and is not is where to put the rigid structure. The obvious approach, and the one the program started with, is to mount a wing on the pilot's back: an aircraft worn like a pack, with its own spars, its own pivots, and pyrotechnic cutters to sever it before the reserve deploys. That architecture is heavy, aerodynamically dirty, and stacks a rigid airframe on top of a non-rigid human in a way that fights the body at every interface. It was abandoned partway through the program when the analysis stopped supporting it, and the replacement inverts the premise. The pilot is not payload sitting beneath a wing. The pilot is the fuselage.

In the current architecture the leading-edge spar is a carbon boom hinged at the spine yoke at the shoulder and running out along the arm to the wrist. The trailing-edge spar mirrors it along the leg, hip to ankle. The pilot's limbs are not passengers inside the wing; they are the wing's central load path and its primary control surfaces. The body sits inside the airfoil thickness rather than slung below it, the structure enclosed by a Dyneema-bearing skin at roughly fifty grams per square meter, the whole thing modeled as a continuous cambered section from tip to tip. This is paraglider thinking applied to a rigid frame, not exposed-spar aircraft thinking. The full span beyond the reach of human bone comes from three-stage telescoping carbon tubes that fire out of the wrist and ankle hubs on compressed carbon dioxide. The pilot is modeled as a fixed-proportion rigid skeleton, upper arm at thirty-two centimeters, forearm at twenty-seven, the bones treated as members that never stretch. The span problem is solved by the telescopes, not by asking the body to do something it cannot.

The payoff of putting the spars on the limbs is that control authority becomes native to a wingsuit pilot's existing instincts. Shoulder and hip rotation against a locked spar maps onto the body-flying a wingsuit jumper already knows. There is no translation layer between intent and surface. That is the kind of property you want to fall out of the architecture for free, and here it does.

The wing got smaller because of where it lands

The single decision that frees up the rest of the design is also the least intuitive, and it is a decision about the end of the flight rather than the middle of it. An aircraft you land on has to land slowly, which means it has to stall slowly, which means it has to be large. A low stall speed is expensive in wing area, and wing area is the enemy of a structure that has to telescope out of a wrist. The original sizing carried an eight point four square meter wing and a best-glide speed around twenty-five meters per second, and it required a telescoping boom on the order of three and a half meters to make the span work. That boom is not feasible as a thing that fires out of a hub on a human arm.

The resolution is that MANTA does not land on the wing. It lands under a reserve canopy. Once you accept that the flight ends with the rigid structure retracted or pivoted clear and the pilot descending on a parachute, the low-stall-speed requirement disappears, because nothing about the touchdown depends on the wing's stall behavior. The wing is now free to be smaller. Resized to six point five square meters and six point three meters of span, the best-glide speed drops to somewhere between sixteen and eighteen meters per second depending on the pilot's mass, and the boom that has to deliver the span shrinks to roughly two point four meters, which is a thing that can actually be built into a telescoping carbon assembly. The performance number that looks like a loss, glide speed falling from twenty-five to eighteen, is the thing that makes the rest of the vehicle constructible. The constraint at the bottom of the flight propagates all the way up to the geometry, and the geometry is held as a single source of truth from which the aerodynamic, structural, and CAD pipelines all read. Change the planform in one file and the glide polar, the mass budget, and the deployment animation all move together. There is no version of the wing that exists in the renders but not in the structural analysis.

Half a second is the easy part; ten milliseconds is the hard part

Deployment is fast, and the speed is not the difficult constraint. The sequence runs in four phases. The pilot extends from a tuck and pneumatic yokes assist and lock the spars at their deployed sweep, about three tenths of a second. Carbon dioxide fires and the telescoping tips snap out of the wrist and ankle hubs, about a tenth. Bistable carbon tape-spring ribs, nine per side, coiled at the spar in the stowed state, snap to their open shape passively as the extended spars pass them, about a twentieth. The skin tensions across the deployed frame and the flight control system captures trimmed glide. The ribs are worth pausing on, because they deploy on stored elastic strain energy alone, no actuator, no latch, no power. A tape spring is bistable; it wants to be coiled or it wants to be straight, and it snaps between the two. Removing power from a deployment step removes a failure mode from it.

The hard constraint is symmetry. The dominant way this vehicle kills its pilot is asymmetric deployment: the left wing locking meaningfully before the right. A timing mismatch between the two sides produces a yaw and roll moment, and if the mismatch is large enough the moment exceeds the pilot's control authority and rolls the vehicle into a bank it cannot recover from before it runs out of altitude. The analysis puts the allowable mismatch at ten milliseconds at three sigma. Ten milliseconds is a demanding budget for a pneumatic system, and the first instinct, a single carbon-dioxide manifold feeding both sides through matched orifices, does not close it. Cartridge temperature and fill variance, manifold pressure-drop imbalance, spar friction hysteresis, and rib-snap dispersion all stack up past the budget. The closure requires active regulation: a solenoid on each side metering flow independently, with six spar-lock microswitches and a left-right timing monitor watching the event happen. If the two sides drift past ten milliseconds, the state machine does not try to fly through it. It aborts to the reserve. The failure is detected and converted into a survivable outcome rather than allowed to escalate into an uncommanded roll. That is the difference between a system that has a dangerous failure mode and a system that has a dangerous failure mode it can see.

The structural margin depends on a control law, which is the uncomfortable part

The coupling that makes MANTA genuinely hard, and genuinely interesting, is between the structure, the flight control system, and the fact that the largest single mass in the vehicle is alive and moves on its own. The pilot is about seventy-nine percent of the all-up mass, and the pilot is not rigid. A fifty-millimeter shift of the head and torso moves the vehicle's center of gravity by roughly three and a quarter percent of mean aerodynamic chord. The static margin at the design lift coefficient is about five point four percent of chord, and it falls below four percent near stall. Set those two numbers next to each other and the problem is visible immediately. A modest, involuntary posture shift by the pilot is comparable in magnitude to the entire static stability margin of the aircraft. The occupant can destabilize the airframe by flinching.

The mitigation is an angle-of-attack limiter in the inner control loop, backed by body-rate damping, running on redundant flight computers cross-checking their state estimates at four hundred hertz with a mechanical reversion path if both lose power. None of that is unusual for a fly-by-wire aircraft. What is unusual, and what the program is explicit about, is that the limiter is not treated as a feature. It is treated as a structural assumption. The spars are sized on the premise that the limiter holds the angle of attack inside a bound, which caps the aerodynamic load the structure ever sees. If the limiter fails, the assumption under the spar sizing fails with it, and the load case the structure was never built for becomes reachable. The control law is load-bearing in the literal sense. This is the kind of cross-domain dependency that conventional design tries to avoid, because it means the structures team cannot sign off on the spar without trusting the controls team's envelope protection, and the controls team cannot relax the limiter without invalidating the structure. The two disciplines are welded together by the physics of a soft, heavy, mobile occupant. You cannot decouple them, so the design does not pretend to.

The standard is whether a coroner would accept the analysis

The program states its bar in one sentence: would this analysis hold up if a coroner's office asked for it. That is the right standard for a vehicle a human wears off a cliff, and it shows up in how the work is organized rather than only in how it is described. Every number in the design brief traces to a script or a document. The aerodynamics run through a custom Weissinger lifting-line solver with the geometry read from the same source file the renders use. The spar is sized by an explicit cantilever bending calculation at a three-g limit load, which is what forced the front spar from a forty-millimeter outer diameter up to seventy-three millimeters with a two-and-a-half-millimeter wall when the original tube failed the check. The failure-mode analysis is not a compliance table filled in after the fact; it is the thing that produced the architecture. Active per-side gas regulation exists because the symmetry analysis demanded it. The mechanical reserve jettison, latched and reversible, replaced pyrotechnic spar cutters because removing an ignition event removes a failure mode. The whole analysis suite runs as a test target, and the current state is twenty-seven of twenty-seven passing across aerodynamics, structures, and deployment.

The most honest moment in the program is the one where the architecture was found to be wrong. The aircraft-on-the-back concept was not abandoned because someone preferred the alternative. It was abandoned because the integrated analysis stopped supporting it, and the wingsuit-extension architecture was the thing the constraints actually pointed at once the pilot was treated as structure and the landing was moved onto a reserve. The CAD, the viewer, and the documents were rebuilt around the corrected concept rather than patched to hide the reversal. That is what real engineering looks like from the inside. The evidence overturns the assumption, and the assumption loses.

What remains is a wing that is mostly forced. The pilot is the fuselage because mounting a separate airframe fights the body. The wing is small because it lands under a reserve and a small wing is the only one that can telescope out of a wrist. The deployment is actively regulated because ten milliseconds will not close any other way. The angle-of-attack limiter is mandatory because the occupant outweighs the aircraft and will not hold still. Each of those is a place where the design space narrowed to one until a defensible answer was the only answer left. A wing that falls out of its own constraints is the kind of thing that is either impossible or close to inevitable, and the only way to find out which is to build the test article and let the gas fire.