Robots That Will Cause A Buzz

A team of researchers at the University of California, Berkeley hopes its invention will create a little buzz and a lot of flap.

The mix of engineers, biologists and others has spent the past four years developing a tiny robot, called the Micromechanical Flying Insect, that members say will one day fly like a fly.

The project is among a handful aiming to engineer devices that can soar, dart and hover on gossamer wings that flap with a rhythm and precision otherwise found only in nature.

The projects are taking different paths, but the goal is the same: churn out tiny devices as nimble as they are small to surreptitiously spy on enemy troops, explore the surface of Mars or safely monitor dangerous chemical spills.

The Pentagon's Defense Advanced Research Projects Agency is funding much of the work because of its potential application in both reconnaissance and surveillance.

In recent years, scientists at Berkeley and elsewhere have made huge strides in understanding the unsteady aerodynamics that allow insects and the smallest of birds to fly. The challenge is now to apply that knowledge to the design of devices that, at least at Berkeley, mimic the size, weight, power and — above all — aerodynamic elegance of a fly.

"What we're targeting is the blowfly, how it specs out," said Tim Sands, a professor of materials science and engineering.

Lest anyone scoff, Sands and his colleagues point out that a fly can lift its own weight, turn more quickly than any fighter jet, zip about even on torn wings — and cap it all off by landing on the ceiling.

"Insects," said Berkeley's Ron Fearing, "have tremendous maneuverability."

In a cluttered campus lab, the professor of electrical engineering and computer sciences uses tweezers to pick up a prototype of the mechanical insect. The robot is a flyweight contender for the title of most ambitious of all the flapping robots, generically called ornithopters, entomopters or micro air vehicles. It has yet to fly.

The Berkeley device is being developed under a five-year, roughly $2.5 million contract. That's pricey for something best described in pocket-change terms.

It takes about a dime's worth of raw materials, including stainless steel that must be folded under a microscope, to build one of the robots. A single penny weighs more than two dozen of the devices. And each boasts a wingspan that matches the diameter of a quarter.

Officials envision soldiers deploying the robotic insects in battle, using them to snoop as only a fly on the wall can.

"It takes an individual and extends their sensory capabilities like a periscope — but it flies independently," said Roy Kornbluh, an engineer at SRI International in Menlo Park. Along with DARPA, the firm has funded development at the University of Toronto of another flapper, a four-winged robot called "Mentor."

During a February flight, the device became the first ornithopter to successfully hover, doing so with the agility of a hummingbird. Mentor is about one foot across and weighs one pound; researchers hope eventually to shrink it down to hummingbird size and weight.

As difficult as flapping flight is to ace, researchers remain enchanted by it because it makes for miniature flying machines that don't gobble large amounts of power.

"Flapping is much more aerodynamically efficient at small sizes, rather than conventional aerodynamics," said Michael Dickinson, a professor of integrative biology at Berkeley and a pioneer in understanding insect flight.

Building wings that flap is one thing, but endowing a robot with enough smarts to control that flapping enough to sustain flight remains difficult, if not impossible.

"The good news is we know what the wings need to do. The bad news is we don't know how to do it," Fearing said.

Consider the fruit fly, Dickinson says. It beats its wings 200 times a second, flapping and rotating them on each stroke in a complicated orchestration that relies on three distinct mechanisms to provide it lift.

In just eight strokes and 40 milliseconds, a fruit fly can make a mid-air U-turn. Fearing estimates that to copy that level of control, the Berkeley bug would have about a three-stroke margin of error. Mistime the fourth, and the fly goes into a death spiral, he said.

Robert Michelson, principal research engineer at the Georgia Institute of Technology Research Institute, said it's too difficult to build a robot that relies solely on modulating its flapping wings for stability and control. Even the Mentor uses four tail-like fins to direct the downwash of its flapping wings to remain aloft.

Michelson said he is developing a flapping robot, called the entomopter, that will use bursts of gas, a byproduct of the device's chemical propulsion system, to adjust the amount of lift provided by each of the robot's twin sets of wings.

"Until we can do things as well as you find them in Creation, you have to go to alternate techniques," Michelson said of his device, which NASA is eyeing for use on Mars.

Size is also a problem: members of the team behind the Mentor said they opted to get their robot flying before shrinking it. The Berkeley team has taken the opposite approach — one that others said may prove overly ambitious.

Michael Goldfarb, an associate professor of mechanical engineering at Vanderbilt University, said limitations in battery and artificial muscle technology will keep the tiniest ornithopters grounded.

Goldfarb's own efforts to build a flapping robot with a six-inch wingspan were unsuccessful.

"Our conclusion to that study was it's not doable with state-of-the-art technology," Goldfarb said.

As it works to get its fly to take wing, the Berkeley team acknowledges it has set its sights high.

"It's a little bit of a moonshot," Dickinson said.