Insect WingA Report on Robotic Insects
by Andrew Holland
© 2000 Australian Broadcasting Corporation
At last one of the longest running scientific mysteries has been solved. Researchers in San Francisco, USA have finally discovered how insects fly.
According to conventional aerodynamic theory the short stubby wings and large body of an insect shouldn't ever get off the ground. But insects, and flies in particular, are the most skillful fliers on the planet.
The breakthrough has come from a biologist with a bent for high tech gadgetry. With the help of giant mechanical wings flapping in a huge tank of oil he has discovered the aerodynamic tricks that make flies top guns. Now that one of natures best kept secrets has been cracked, the next step will be to build a robot fly.
NARR: Deep in the basement of this building a strange creation has come to life. Meet 'Robofly'. With a flap of his wings, Robofly and his creators have solved one of the longest running scientific mysteries - 'how insects fly'. Aerodynamic theory has been changed forever, and remarkably, the development of a flying robot insect is a real possibility. The brain behind Robofly is Michael Dickinson. A neurobiologist by training, his breakthrough in bug aerodynamics came almost by chance. He initially wanted to learn why flies have tiny sensors that tell them how much their wings rotate.
NARR: scientists had a very difficult time explaining how an insect could keep itself in the air at all and so in order to really tackle the problem of wing rotation I had to go back to square one and start to work on the question of insect aerodynamics. So why not just look at how planes and birds stay aloft? Their flight can be explained through a theory called "steady state aerodynamics". The shape and slight angle of their wings makes air travels faster over the top surface than the bottom. This creates lower pressure above the wing, generating lift. But if you apply aircraft aerodynamic theory to the stubby wings and large body of an insect, the theory crashes. It predicts that insects shouldn't be able to get off the ground.
Michael Dickinson: Well the basic problem is that an insect unlike an aeroplane is constantly moving its wing back and forth so most insects use a horizontal stroke plane so they're not actually flapping their wings up and down as a bird would but they're flapping their wings back and forth and so the wing is actually only moving a little, a short distance before it rotates and moves back.
NARR: A few years ago researchers in the United Kingdom thought they'd found the answer to insect flight. After converting a wind tunnel into a 'moth disco', they discovered an aerodynamic effect created by the steep angle of the insect's wing.
Michael Dickinson: What's different about a fly and other insects compared to aircraft is that they use a really high angled attack during each stroke - a much higher angle of attack than is used by conventional aircraft and that generates a large leading edge vortex, a large swirling vortex on the top surface of the wing that generates very low pressure and consequently pulls the wing upward.
NARR: Amid great excitement the aerodynamic discovery was dubbed "delayed stall". But when subsequent sums were done, it became apparent that delayed stall wasn't the complete answer. It couldn't provide enough lift force on it's own. There must be even more aerodynamic tricks at work. Michael Dickinson figured researchers needed a new trick of their own.
Michael Dickinson: Unfortunately non steady state aerodynamics is so complex that even super computers have a difficult time predicting the forces that are generated by an insect wing flapping back and forth and that's why Charlie Ellington and myself and other researchers have used the strategy of dynamic scaling actually building a large insect that flaps very slowly in order to study the forces generated by an insect.
NARR: He decided to build a set of robotic insect wings based on a hovering fruit fly, drosophila. To accurately measure the forces produced by the real insect - here slowed down to one hundredth of normal speed - Michael copied its wing shape and complex wing rotation. His creation, Robofly, had to be scaled up to allow for the positioning of force sensors at the base of a wing. And to keep the physics the same, he made the wings flap slowly in a huge tank of oil.
Michael Dickinson: we can't use air because to a small insect the air would feel very thick and syrupy so we use high viscosity mineral oil, we have two metric tonnes of it.
NARR: Up till now researchers had relied on smoke patterns produced in a wind tunnel. But Robofly enabled Michael to directly measure the forces produced by flapping insect wings.He then only needed some simple mathematics to calculate the all important lift force. Sure enough Michael's team found that 'delayed stall' played a major role, but they also measured two previously unexplained aerodynamic effects. At the end of each wing stroke they found a lift force created by the rotation of the wing.
Michael Dickinson: This is the force that's very similar to the force on a tennis ball that has backspin. As a tennis ball spins it pulls air over the top faster than the bottom and as a consequence higher velocity means lower pressure the ball is sucked upwards and effectively the wing is being sucked upwards as it rotates.
NARR: And at the beginning of each wing stroke Robofly revealed an even more surprising source of lift.
Michael Dickinson: This is due to a different mechanism, a rather exciting mechanism that we call 'Wake Capture' and this comes about because an insect flaps its wings back and forth and as a consequence the wing is always passing through the wake of the previous stroke and it's able to actually extract energy from the wake and this makes the wing beat rather efficient and it's a wonderful way of - that the insect can generate forces above and beyond what you would predict from conventional aerodynamics.
NARR: Now that Michael Dickinson has cracked one of nature's best kept secrets, his research lab has taken off. He's resuming his studies of brain and muscle interaction in flies, but he also wants to learn more about insect aerodynamics. While Robofly duplicated a hovering insect, Michael now plans to let his creation travel.
Michael Dickinson: we're already designing the bride of robofly if you will, a new test platform in which the entire flapping device will actually translate through again a tank of mineral oil and so this will allow us to study how the aerodynamics change when the animal is actually moving through the fluid.
NARR: And amazingly his discovery has helped launch a project that aims to build a flying robot insect. An artificial fly could search for the victims of an earthquake or sniff out the smallest gas leak. The applications are still being dreamt up. But the robots design is well underway.
Part 2 : Robot insects coming soon
A new field of technology called biomimetics aims to create machines that mimic living creatures. With funds from the US military, there are projects under way to build robots that swim like a fish, dive like a lobster, and. fly like an insect. The first robot fly is due to be hovering in the still air of a laboratory by 2003.
To meet that deadline researchers will have to overcome a few technical hurdles - like the need for revolutionary batteries, brand new materials, and micro-computer systems that are the stuff of dreams. Sounds like its a long shot? Not so. The flying robot insect is already taking shape.
NARR: This is the design of a not so secret weapon for the new millennium- a micro-robotic flying insect, just 25 millimetres from wing-tip to wing-tip. This is no joke- the project has the backing of the US Navy and the US Defence Advanced Research Agency. Here at the University of California, Berkeley, researchers are planning to test-fly a hovering robot fly within 3 years. At this early stage "Louie" is harmless and flightless, but the sky really is the limit. A robot fly's ability to weave through the most intricate obstacles would be a blessing for search and rescue teams. Simple sensors could hone in on trapped people or warn of a gas leak.
Michael Dickinson: I can imagine every fire hall having a little jar of robotic insects that when the rescuers come to a collapsed building they could just start to sprinkle over the rubble and these things could crawl and fly and just try to find, find people and then just simply beam back their location.
Ron Fearing: There's no reason why you couldn't you know as Michael Dickinson said just have a jar full of these and you know throw out five hundred and if 20% find something of interest that's fine, don't worry about each individual one.
NARR: Why copy a fly? Well, it seems an aeroplane wing won't work at an insects scale. A micro-helicopter may be possible, but the bet is a robot fly would have the edge in manoeuvrability. Of course there are no off-the-shelf robot insects. Before test flights can begin the Berkeley team will need to develop a host of new technologies.
NARR: Neurobiologist Michael Dickinson has already cracked one of the toughest challenges. As shown on Quantum last week, his pair of giant fly wings have finally revealed how insects stay in the air. Now Michael's new aerodynamic theory is being used to develop artificial wing mechanisms.
Ron Fearing: After a lot of head scratching sort of as a joint brain storming we came up with the idea of taking two simple mechanisms each one which flaps up and down and combine that with a flexible wing structure so (late in?) the way this works is we have two individual pieces each one goes up and down independently and if the two go together the wing just flaps. If the two go out of phase which each other then we can get equivalent of a wing rotation.
NARR: But pulleys, gears and cams would be too cumbersome at bugs scale. So Project Leader Ron Fearing turned to a fly for inspiration. It's wing and hinge mechanism, or thorax, is essentially a single structure- a kind of intricate piece of origami. The thorax translates tiny muscle movements into large wing motions. So beginning with an ultra thin sheet of steel, a laser cutter, and some very fiddly folding, Ron came up with his own origami wing mechanism.
Ron Fearing: So this is about a five times size and then we need to go another, another five times below this but this demonstrates the basic principles. So by folding up this structure we get again the four bar and can control this individual wing spar we have a flexible wing and that will give us the wing rotation and the wing up and down.
NARR: But will it ever fly? The team have just begun testing their design in a wind tunnel.
NARR: So far so good. Comparisons with the forces generated by a real fly show that the wing concept is on the right track. Of course getting an artificial wing flapping won't automatically produce a top gun robot. That'll require rapid control systems just like a real fly.
Michael Dickinson: Well I think flies are the most sophisticated flying devices on the planet.
Michael Dickinson: Flies have turned their hind wing into a gyroscope so they actually have an on board gyroscope that is constantly monitoring the rotation of the body during flight and that information is being used on a moment by moment basis to control wing kinematics that keep the animal flying stably. So how do you study the flight control systems of a creature that buzzes all over the place? Michael Dickinson has come up with a kind of virtual reality chamber he calls the "Flight arena". A fly is glued in front of an array of individual diode lights. The remarkable thing about this experiment is that the fly actually controls the pattern of lights as it changes the frequency of it's wing beats. For some reason they prefer to fly towards vertical lines.
Michael Dickinson: We allow the flies to play a little video game. We can measure in real time how they move their wings and what aerodynamic forces that they generate and those signals can be used to control the fly's visual world so that if it made a turning force to move in one direction we have a visual panorama that moves appropriately in the opposite direction so the fly can sort of steer itself through a little virtual landscape but since it's stuck in one spot, since we fooled it into thinking that it's flying around we can more easily study its flight system.
NARR: What's unique here is that the device allows researchers to get inside the loop connecting the flies eyes, it's stabilising gyroscope, and wing mechanism.
NARR: Once the robot is flying stably, it will have to work out where it needs to go. To solve the mystery of how flies locate a target, Michael's team has developed another ingenious experiment. A fruit fly is released inside a giant cylinder dubbed the"Flyarama". Stereo cameras track the fly's movement in three dimensions.
Michael Dickinson: It enables us to look at the three dimensional complexity of the flight trajectory. So this a bird's eye view if you will of the flight trajectory but we can rotate things a bit so you can get a little bit more of the 3D complexity. You can see that the fly is not just navigating in the horizontal plane but it's actually moving up and down in the vertical direction as well.
NARR: A startling pattern emerged when Michael analysed the three dimensional flight paths - flies like to make right-angled turns.
Michael Dickinson: "Here's a good example here of a straight flight path and a fly very rapidly turns at about 90 degrees to fly off in another direction and if you average these secotic turns over many individuals and many turns of a given individual you come up with almost exactly 90 degrees and what the flies will actually do as they fly around searching for something is make about four or five 90 degree turns in one direction, then they make about four or five 90 degree turns in the other direction and this seems to be their default flight mode where they're actually scanning the world for interesting objects."
NARR: The strategy clearly works - fruit flies have no trouble finding a rotten banana. Although it'll be a while before an artificial insect is able to do anything as useful. But by 2002, the first generation of robotic flies could be hovering in the still air of a laboratory-if all goes well.
Ron Fearing: We don't see any fundamental reason from an engineering point or a scientific point here why this couldn't work. If we solved the power problems, if we solve the structure problems, solve the control problems. It looks feasible.
Associate Professor of Integrative Biology
Electrical Engineering and Computer Science
Senior Science Writer
Return to Aerobotics Index