Monday, August 3, 2009




In 1804 the English aviation pioneer George Cayley installed a bizarre machine at the top of his staircase. He attached wings of various shapes to a whirling arm atop the device, and as it spun the wings would either climb or descend depending on their ability to generate lift. This helped Cayley to develop the aerodynamic theories that led to his successful manned glider flights, and ultimately to the Wright brothers' powered aircraft.

More than two centuries later, a whirling arm is once again being used to prepare the next revolution in flight technology: micro-aircraft that harness the complex aerodynamics and navigation techniques of insects. In his lab at the University of California, Berkeley, microsystems engineer Ronald Fearing fixes each new version of the mechanical insect he is developing to the tip of a 30-centimetre free-spinning arm he calls a "flight mill". Like Cayley's machine, this allows him to measure how much lift his creation can generate, and to test different ways of controlling it.


Mechanical insects could prove far more manoeuvrable than micro-sized versions of conventional aircraft or helicopters. The insect-like craft could fly unobtrusively around buildings, zipping into open windows, for example. When equipped with different kinds of sensors, they could be used as miniature spy drones, security guards and pollution monitors.

The military in particular are interested. The Pentagon's Defense Advanced Research Projects Agency is developing four flying "robobugs", weighing up to 10 grams each and with wingspans of up to 7.5 centimetres. One of the two companies developing the craft for DARPA - Aerovironment, based in Monrovia, California - aims to have a "rough demonstrator" flying by the middle of 2008.

It is challenging work. If micro-aircraft like Fearing's are ever to fly, they will not only need to generate lift in a similar way to insects, but also mimic the way bugs sense their environment to allow them to maintain stability and land safely. Recent developments in wing mechanics and control systems mean that researchers are now getting close.


The first hurdle for engineers like Fearing is to develop mechanisms that will generate enough lift. Insects do this by rapidly beating their wings down and forward, and then rotating them back and upward (see "Moth in a wind tunnel"). At last week's Society for Experimental Biology meeting in Glasgow, UK, a host of new robotic insect-wing designs and flapping mechanisms were on display. Andrew Conn at the University of Bristol in the UK unveiled a hummingbird-sized wing mechanism driven by a pair of motorised aluminium cranks that reproduce a typical insect wingbeat: one beats the 7.5-centimetre wing up and down, while the other rotates it (see Photo, above). Unlike previous mechanisms, says Conn, the current design's wing motion is adjustable and should allow more manoeuvrability in the air.

However, the team, which is being funded by the UK government's Defence Science and Technology Laboratory, has found that friction in the mechanism is slowing the wing's beating. The device is also currently too heavy to take off, so the researchers plan to replace as much metal as possible with carbon fibre. "We'll probably need to halve our weight and at least triple our lift," says Conn.

These problems come as no surprise to the entomologists at Michael Dickinson's lab at the California Institute of Technology in Pasadena, where they study fly and honeybee wing dynamics. Anyone attempting to mimic insect wing motion using such complex machined gearing may be wasting their time, they say. As the Bristol team is finding, friction dominates at such small scales, so micro-sized versions of conventional gears and pulleys can sometimes seize up. Dickinson's team reckons success is much more likely to come by emulating the way an insect uses muscles to flex its whole thorax, which in turn moves the wings.

This is the approach being followed by Fearing, who has worked closely with Dickinson. For his 0.1-gram Micromechanical Flying Insect (MFI), he has gone for a more insect-like approach. The prototype comprises a 2-centimetre-wide carbon fibre "thorax" with 4-millimetre polyester and carbon fibre wings on either side. To move the wings, two piezoelectric actuators move a concertina-like carbon fibre structure incorporating 15 polyester joints. As the piezoelectric crystals expand and contract they flex the joints back and forth. The flexing thorax is attached to each wing by a hinge to drive the down-and-forward, up-and-rotate-back wingbeat characteristic of insects (4MB .avi video).



Until March 2006, a one-winged MFI was buzzing around Fearing's flight mill, creating just 500 micronewtons of lift. That meant two wings would provide the 1000 micronewtons needed to get it airborne - but not enough to allow sensing equipment to be attached to the device. To boost the lift, Fearing once again turned to Dickinson. In 2005 the Caltech team had demonstrated how honeybees' unique wing motion allows them to generate enough lift to fly, despite their heavy bodies and short wing beat . So Fearing switched his insect flight model from a fly to a bee, increasing the MFI's wing stroke from 170 beats per second to 275, and reducing the angle through which the wing moves up and down from 70 to 60 degrees. This has tripled its lift . "The critical thing that we have shown is that we have enough lift to take off."



Generating lift is only half the problem, though. The micro-aircraft will also need precision 3D flight control once in the air. Remote control is one option, but they will be more useful if they can be autonomous, and for this they will need to mimic another part of the insect's repertoire.
The way it changes direction, avoids walls and moves indoors is just like a housefly

Insects navigate by monitoring the way surfaces around them - most obviously the ground - sweep backwards in their field of vision as they fly forward. This "optical flow" provides cues about their airspeed and height that are crucial for landing safely, avoiding obstacles and navigation in general. The bug can be sure it is hovering, for instance, when it senses no optical flow. When flying forwards at a steady speed, the insect "knows" that if optical flow decreases, its altitude must have increased. When it comes in to land, a bug slows down safely as it approaches the ground by ensuring the flow rate stays steady. Fearing plans to recreate this ability to sense optical flow by adding a fisheye lens above a light-sensing chip that will feed optical flow data to the machine's microchip brain (see Photo, left).

At the Swiss Federal Institute of Technology in Lausanne, flight researcher Dario Floreano is already testing optical-flow sensing software on a miniature propeller-driven aircraft. Dubbed Microflyer, it tracks the position of features on the ground and walls (13.6MB .mov video) using two cameras scanning below and ahead. The aircraft may not look much like a bug but it certainly flies like one, he says. "The way that it behaves, changes direction, avoids walls and moves indoors is just like the way a housefly moves," says Floreano.



If robotic insects do fly, Fearing believes they will quickly become cheap and commonplace. "Something that weighs less than a tenth of a gram will sell for less than a buck," he says.
Moth in a wind tunnel

It was not until the late 1990s that researchers finally discovered how insects fly.

Until then, aerodynamic theory could not explain how insects' small wings create enough lift to get the creatures airborne. As conventional wisdom had it, lift is a result of lower-pressure air flowing over the top of a wing, thanks to the differing curvature of its upper and lower surfaces and the wing's angle relative to the airflow. Yet insects somehow produce up to three times more lift than this model suggests. There had to be something else going on.

In December 1996, a team led by Charles Ellington, a zoologist at the University of Cambridge, found out what it is. Using high-speed video in a wind tunnel, they filmed smoke trails as they wisped over the wings of a tethered hawkmoth, which has a 10-centimetre wingspan. This showed that the insect's complex wing motion - beating down and forward, then rotating back and upward - was generating tiny whirlwinds that moved along the leading edge of each wing  

After further experiments with a 10-times life-size mechanical model of the moth, they found that these vortices were being created on the downstroke and producing a low-pressure area above the wing that can give up to 50 per cent more lift than is needed to loft the creature.

No comments:

Post a Comment