The Berkeley team is one of about a dozen groups of engineers and biologists who are exploring the final frontier of flight: micro air vehicles (MAVs). By merging the aerodynamics of insects with GPS navigation and molecular electronics, they hope to initially create an arsenal of tiny reconnaissance tools. When perfected, Fearing's stainless steel and Mylar robot flies will be able to flap their way into the most secret places on Earth-the bunkers where Saddam Hussein plans his genocidal campaigns, and where Chinese spymasters plot their raids on America's nuclear weapons laboratories.

After this first generation of MAVs have proven themselves as effective spies, they will become armed and dangerous. Alan H. Epstein, of Massachusetts Institute of Technology (MIT), recently described one MAV combat scenario in the technical journal Aerospace America. He envisions GPS-guided MAVs landing on structurally critical points along bridges deep in enemy territory. Each MAV would carry a small piece of shaped-charge plastique. Responding to a command transmitted from half a world away, the MAVs would explode in sequence, bringing down the bridge with only one-hundredth of the amount of explosives required by a pinpoint-accurate smart bomb.

Looking to warfare in 2020 and beyond, some military strategists envision swarms of robot flies fluttering onto battlefields. Scout flies, equipped with miniature cameras, would do the work of reconnaissance teams by eavesdropping on tactical communications and sending back real-time video of enemy positions. Sniper flies would seek out field commanders, recognizing them by the iris patterns of their eyes. Then, they would become the 21st century incarnation of the tribesman's poison dart as they hurled themselves into the carotid arteries of their targets.

Meanwhile, titanium-tipped robot flies too small to register on radar screens would gather in the weeds at the end of enemy runways. Then, rising as a swarm, they would allow themselves to be sucked into jet engine air intakes. The MAVs titanium bodies would fracture the whirling turbine blades and send a rain of red-hot fragments through thousands of pounds of jet fuel and ammunition. In Pentagon parlance, MAVs have the potential to become the ultimate force multiplier.

As futuristic as these scenarios may seem, the Pentagon's four premier research funding organizations-the Army Research Office, the Office of Naval Research, the Air Force Office of Scientific Research and the Defense Advanced Research Projects Agency (DARPA)-are taking them seriously. Together, they have put upward of $50 million on the table to create the flapping-wing airframes, microscopic jet engines and molecule-size avionics packages needed to make MAVs a reality.

Building MAVs is more complicated than scaling down large aircraft. "Engineers say they can prove that a bumblebee can't fly," says Berkeley's Michael Dickinson. "And if you apply the theory of fixed-wing aircraft to insects, you do calculate they can't fly. You have to use something different." This is where biologists like Dickinson come in, to develop a general theory of insect flight that can be mechanically duplicated with tiny robots.

Creating lift, the force that keeps flying machines aloft, and maintaining control once airborne, involves a host of aerodynamic factors. The most important characteristics can be summarized in what aerodynamics experts call a Reynolds Number. Large, fast-flying commercial aircraft-a Boeing 747, for example-have high Reynolds Numbers, well exceeding 100 million, says Thomas J. Mueller, Roth-Gibson Professor of Aerospace and Mechanical Engineering at the University of Notre Dame in Indiana. As planes shrink and move more slowly, their associated Reynolds Numbers decline. A fixed-wing, 6-in. MAV cruising at 30 mph might have a Reynolds Number of about 130,000, says Mueller. As a consequence, it becomes more difficult to control. Reducing an aircraft to the size of a large insect reduces the Reynolds Number to below 20,000. At a number this low, fixed-wing aircraft need turning radii that would be too large to make the kinds of sharp right-angle turns needed to navigate tight spaces like ventilation ducts.

To be a fly on the wall, MAVs will have to fly like flies-in other words, flap their wings.

"The size and the Reynolds Numbers of MAVs correspond to very small birds," says Mueller. "We have very little information on the performance of these airfoils and wing shapes, but there has been a long history of natural flight studies with insects and small birds that may be helpful." In June, Mueller invited the leading experts on bird and insect flight from around the world to meet with MAV designers for a conference titled "Fixed, Flapping and Rotary Wing Vehicles at Very Low Reynolds Numbers." One of the most important messages for MAV designers was that even though insects flap their wings, they don't fly anything at all like birds.

"You can't fly like a bird if you're the size of an insect," says Berkeley's Dickinson. "You have to fundamentally rethink the problem." Bird wings might not appear to have much in common with plane wings, but the equations that describe flight are fundamentally the same for passenger planes or passenger pigeons. "Steady-state aerodynamics of airplanes works well for birds," says Dickinson. "If you treat a bird wing like an airplane wing and at any given time calculate the speed and lift, then sum it up over the entire stroke, it explains how the bird can stay aloft. With insect flight it fails miserably." About a half-dozen efforts at duplicating bug flight are now under way.

Small, flapping-wing MAVs can take advantage of a new type of motor that produces linear rather than rotator motion. It is called a piezoelectric motor and operates like the needle on a turntable, only in reverse. Piezoelectric materials are crystals that produce a tiny current when they are placed under pressure or otherwise mechanically deformed. Some of these materials respond to current by moving. Their high-power density means they are capable of high force output. At the Center for Intelligent Mechatronics at Vanderbilt University in Nashville, Tenn., researchers have successfully applied this theory to build tiny piezoelectric actuators that can flap wings. At Auburn University in Auburn, Ala., researchers have created materials that change flight-control surfaces using the same principle.