In both planes and birds, different wing shapes yield different types of flight. The Global Hawk surveillance drone, like the albatross, has wings that are long, thin, and narrow, ideal for long-distance, low-speed flight. Planes that need to maneuver at high speeds, like the F-16 Fighting Falcon, have stubbier, swept-back wings, which produce enough lift but with less drag. Eagles and hawks, likewise, have shorter wings for greater agility.

But birds and planes control their flight very differently. Conventional airplanes maneuver by means of moving surfaces: flaps and ailerons on the wings, horizontal sections called elevators on the tail, and also the rudder. Birds, on the other hand, can bend, twist, and deform their wings and bodies to turn, change their speed, and adapt to unforeseen conditions such as wind gusts. If planes could do the same, they would have more lift and less drag, gaining agility and consuming less fuel.

We studied the wings of various birds in search of an appropriate model for our aircraft. Hawks and eagles looked promising, but what really grabbed our attention was the pteranodon, a carnivorous pterosaur that soared through the skies more than 75 million years ago. With membranous wings each nearly 5 meters long, this animal was a formidable glider.

But was that the ideal wing shape for us? To find out, we turned to a computer program called Wind. Created by researchers at the Air Force's Arnold Engineering Development Center, NASA's Glenn Research Center, and Boeing, Wind is a computational fluid dynamics program that lets you simulate an airplane's wing—or, more accurately, its airfoil, the shape characterized by its teardrop cross section—under varying conditions.

From our studies, we needed a thin and slightly curved airfoil that would approximate that of a pteranodon wing. Consulting databases that catalog thousands of existing airfoils, we found a few with the right profile, bearing the code names Selig 1091, Selig 1223, and Eppler 378. Using Wind, we simulated each airfoil for high-altitude flight at relatively low velocities of up to 64.6 meters per second, or Mach 0.19. The analysis produced two parameters that are essential in any aircraft design: the lift coefficient (which, as the name implies, is a measure of the wing's ability to push the aircraft upward) and the drag coefficient (a measure of unwanted resistance to motion). These two parameters vary according to the wing's angle of attack, its front-to-back inclination relative to airflow. It's kind of like flying a kite: to get it up in the air, you pull it at an angle to the ground, so as to maximize lift.

When designing an aircraft, then, you study how lift and drag vary as you angle the airfoil. Our initial tests showed that for an angle of attack of 10 degrees, the Selig 1091 has a maximum lift coefficient of 1.5 and a drag coefficient of 0.05—good enough, aerodynamically speaking, for our type of aircraft. But our plane, like a bird, will constantly change how it angles its wings forward, so we had to study our wing's aerodynamics over a wide range of angles.

Before we settle on a final airfoil, we'll also test two- and three-dimensional computer models, as well as actual scale models in a wind tunnel, simulating myriad steady-state and turbulent conditions as the plane flaps, glides, and soars.

Flapping-wing airplanes, sometimes called ornithopters, have fascinated humanity for centuries. Leonardo da Vinci proposed a few such designs, to be powered by a human, but whether they were built remains unknown. More recently, inventors have successfully demonstrated both small and large ornithopters. A team at the University of Toronto, for example, developed an ingenious flapping plane, powered by an internal-combustion engine, that even carries a pilot on board.

Despite such successes, we decided to depart altogether from traditional aircraft design paradigms. In most planes, the wing is a cantilevered framework covered by metal plates that give the wing its aerodynamic shape. We rejected the framework structure in favor of layering together different materials to form a compact, nonhollow body.

Enter the aforementioned ionic polymer-metal composite, or IPMC. Outwardly, it looks like ordinary plastic, but its core is made of perfluorinated sulfonic acid, a compound that works as an ion-exchange membrane: when exposed to an electric field of a few tens of volts per millimeter, it allows water molecules and hydrated ions to migrate across it. This flow of water and ions creates internal forces that make one side of the sheet expand and the other contract, resulting in a bending motion. The deformation is proportional to the electric field's strength, and once the field is removed, the sheet returns to its original shape.

The solid-state aircraft will have wing-shaped sheets of the material sandwiched between two metal grids: an anode grid on the bottom and a cathode grid on top, each containing thousands or even tens of thousands of electrodes. A computer control system will supply voltage to the electrodes. By applying different voltage levels to different portions of the sheet, we can make it change its shape to flap. Our hope is that with a very fine grid we'll be able to emulate the wings of flying vertebrates.

We're far from that level of control. In our first test, we started small, with two 5-centimeter-long IPMC strips. We then attached control electrodes to each little wing and applied a variable voltage. It was fascinating to see this primitive contraption beating its wings fairly well, even at high flapping rates.