From the issue dated July 2, 1999


A Berkeley Professor Uses Robotic Wings to Explain How Insects Fly

The flitting of a bee from flower to flower, the bloodthirsty hunt of a mosquito on a summer evening, and the maddening meandering of a housefly: Each seems so ordinary, so ubiquitous, that it never occurs to many people to wonder how it's possible.

Not Michael H. Dickinson. The associate professor of integrative biology at the University of California at Berkeley has turned entomology on its collective ear, with research using a pair of robotic insect wings that now offers, for the first time, a comprehensive explanation of how insects fly.

His study, the results of which were published in the June 18 issue of Science, documents three types of wing motions that produce enough lift for insects not only to fly but to perform the aerodynamic acrobatics that are part and parcel of insect life aloft. Besides shedding light on how insects fly, the research could form the basis for designing insect-size flying robots, says Mr. Dickinson.

His test device, known as Robofly, is a pair of Plexiglas wings modeled after the wings of the common laboratory fruit fly Drosophila melanogaster -- but 10 inches long, instead of a small fraction of an inch. The faux wings are controlled by a computer that has been programmed to direct motors to beat them using the same motions that a fruit fly does. The wings are set in a container of mineral oil, which provides the same amount of resistance to Robofly's scaled-up wings as does air to the tiny wings found on real fruit flies.

Sensors attached to Robofly's wings measure the forces on them at each instant, allowing Mr. Dickinson and his collaborators to determine which parts of the wing motion generate the most lift for a flying insect. "Robofly actually measures those forces directly, so you know exactly where in the wing stroke the forces are produced," says Mark A. Willis, an assistant research scientist at the University of Arizona's Arizona Research Laboratories.

He recalls his reaction when he first saw Mr. Dickinson's results with Robofly, at a scientific meeting in January. "I was blown away," Mr. Willis recalls. "It was so cool."

Researchers have long struggled with explaining insect flight. They could not invoke the same aerodynamic forces that are known to keep both birds and fixed-wing aircraft aloft, in which lift is produced by the difference in the speeds with which wind traverses the top and bottom surfaces of a wing. Insect wings are so much smaller than the wings of both planes and birds that, mathematical computations showed, the process simply could not generate enough lift to keep insects airborne.

"You can't make a model 747 and expect it to work when it's the size of a fly," says Mr. Willis.

"It's been a bit of an embarrassment that we couldn't explain it," Mr. Dickinson says of insect flight. "Think if we couldn't explain the physics of an organism walking."

Charles P. Ellington, a reader in zoology at the University of Cambridge, proposed about five years ago that insect flight relied on a phenomenon called "delayed stall," produced when an insect sweeps its wing through the air at a sharp angle. A conventional plane wing at such an angle would lose lift and cause the aircraft to stall, but Mr. Ellington suggested that insects might develop vortexes of air atop their wings that would generate extra lift.

Mr. Ellington's theory generated a lot of excitement. "A few years ago, it looked like we knew the answer about how insects stay in the air," Mr. Dickinson says.

But as researchers examined the theory in detail, he says, they began to realize that delayed stall would provide an insect with only enough lift to prevent it from falling to the ground, but not enough lift for the sophisticated maneuvering that enables many insects to flit about -- and certainly not enough lift to allow insects to carry objects, such as grains of pollen, in flight.

That was a serious drawback, because it contradicted reality. "Most insects can support two times their body weight," says Mr. Dickinson.

Mr. Dickinson's experiments with Robofly found that delayed stall does happen as a fruit fly sweeps its wings up and down. But he found that delayed stall was only a partial explanation, because there are two other mechanisms that also produce lift for the fly. Unlike delayed stall, both occur when the fly rotates its wings between strokes rather than during the strokes themselves.

One, called "rotational circulation," happens when a wing nears the end of a stroke and rotates backward. That rotation pulls some air and creates lift, just as a baseball thrown with topspin will curve toward the ground.

The second previously unknown mechanism, called "wake capture," occurs when a wing's rotation moves the wing through a wake created by its flapping motion a split-second earlier. That phenomenon increases the force generated by the wing, Mr. Dickinson has found.

According to his calculations, those two rotational processes account for 35 per cent of the lift produced by a fly's wings, enough to account for the details of insect flight that delayed stall alone cannot explain.

"They're just incredible," he says of insects' use of their wings. A fly can fine-tune its rotational lifting power in a way that it apparently cannot with delayed stall. Under certain conditions, by adjusting the timing of its wing rotation by as little as 8 per cent, the insect can almost double the amount of lift that its wings generate, says Mr. Dickinson.

"This is basically a unified theory of insect flight," he says. Although different insect species may differ in some details, he says, "we now have the basic palette of mechanisms that can explain flight." Hummingbirds may also use some of the same mechanisms, he adds.

Mr. Willis of Arizona agrees that Mr. Dickinson's advance is significant. "It's a clever and a pretty complete understanding of how the forces are produced," he says. Indeed, he is planning to collaborate with Mr. Dickinson on a project in which Robofly would be reprogrammed to mimic a moth species that Mr. Willis studies.

But Cambridge's Mr. Ellington -- who reviewed Mr. Dickinson's paper for Science and recommended its publication -- is nevertheless a bit restrained. "I wouldn't have gone that far," he says of Mr. Dickinson's claims for a final, comprehensive theory of insect flight. "There's quite a bit of work still to be done."

Noting that even Mr. Dickinson found that most of an insect's lift came from delayed stall, he says he sees Mr. Dickinson's research as "complementary" to his own. The two rotational processes that Mr. Dickinson has identified, he says, simply add to the lift generated by the dynamic-stall process that Mr. Ellington himself found.

"It's all coming together quite nicely now," Mr. Ellington says.

The research could have practical uses in the design of insect-size robots. The Defense Department, for example, is financing research into the development of tiny, flying devices that could be dispatched in swarms to spy on enemy forces. Mr. Dickinson says he has a grant from the Office of Naval Research and the Defense Advanced Research Projects Agency to try to apply the principles learned from Robofly to such devices.

Using data from Robofly, engineers could produce a flying robot that works exactly like a real insect, or Robofly could be used to experiment with variations on rotational patterns found in nature, says Mr. Willis. "This will help us figure out how closely we have to model biological flight to get synthetic flight that works," he says.

For Mr. Dickinson, however, Robofly's primary attraction lies in how it sheds light on the basic scientific question of how insects fly. Insects are the most successful creatures on earth, found across the planet and over hundreds of millions of years, he says. "What made them successful is the development of flight. Anything that gives us a better understanding of how they fly is important from an intellectual perspective."

"Just walk outside and look at what's buzzing around your garbage can," he adds. "These things are just so damn cool."

Copyright © 1999 by The Chronicle of Higher Education
July 2, 1999, Issue
Section: Research & Publishing
Page: A17