CHAPTER SEVEN
Insects Can’t Fly
The fundamental laws of aeronautics,
dynamics, and what ever
must soon convince the unbeliever
that bees were built to such a model,
they scarcely could do more than waddle.
The ratio of their body weight
to wing-span, he could demonstrate,
precluded takeoff, much less flight.
SHEENAGH PUGH,
‘Bumblebees and the Scientific Method’
When Primo Levi asked, ‘Would all the philosophers and all the armies of the world be able to construct this little fly?’ it would never have occurred to him that any army would want to do such a thing. But they do because the fly is a miracle of robotics, turning at right angles on a dust speck, hovering, avoiding obstacles (except the visually deceptive windows) with ease: a wonderfully manoeuvrable little dogfighter with, according to Rafal Zbikowski, an engineer at Cranfield University working on flapping flight, ‘less computational power than a toaster’.
Not so long ago, aeronautics engineers were disdainful of natural flight mechanisms. In The Simple Science of Flight, Professor Henk Tennekes, an early enthusiast, tells how, back in 1969, he dared to talk about bird and insect flight in his courses. His boss informed him: ‘In your class you seem to have talked about geese and swans. I cannot condone that. Our profession – mine, and I trust yours, too – is a branch of engineering. Animals that flap their wings are none of our business. Please restrict yourself to airplane theory.’
Although the Wright brothers studied vultures in flight, aircraft flight owes little to birds or insects, and for a long time conventional theories of flight could not account for the amount of lift generated by insects. The equations said that such creatures should not be able to fly. It is chastening not to know how they do it. If we could fly at twice the speed of sound from London to New York, how is it that we did not understand this alternative form of aerodynamics?
But in patient work over recent decades, biologists in Britain and America have unravelled much of the complexity of insect flight and brought the science to the point where engineers can contemplate making flying vehicles that flap their wings.
But what kind of plane would flap its wings? Clearly, it is not going to be a rival to the 550-seat Airbus A380. The US Defense Advanced Research Projects Agency (Darpa) initiated the research in 1996 with a $35 million development programme to develop ‘micro aerial vehicles (MAVs)’. The idea goes back to 1992 when Darpa conducted a workshop entitled ‘Future Technology-Driven Revolutions In Military Operations’. One of the topics of that workshop was ‘mobile microrobots’. Various studies demonstrated that the concept was feasible, so the call went out. The Darpa project formally ended in 1999; they were, as so often, a little ambitious with the timescale but the work goes on in several laboratories.
The target of the Darpa project was a flying vehicle that could fit into a 15 cm sphere, weigh no more than 140 g, fly for up to 2 hrs and have a range of 10 km, operate in winds of up to 50 kph, and be able to manoeuvre without a remote pilot sitting at a control panel guiding it. This specification does not require or necessarily imply flapping flight, but two teams, led by Ron Fearing at the University of California, Berkeley, who we have already met for his gecko work (see page 85), and Rafal łbikowski at Cranfield University in the UK, have decided to go down the flapping route, and in very different ways: Zbikowski is designing a machine at the 15 cm limit specified by Darpa, while Fearing is working on the much smaller scale of the blowfly (a large housefly with a 25 mm wing span).
Before looking at how insects fly it is worth going over how aeroplanes do it. Despite the innovations of 100 years of aviation -from the first biplane that hopped off the grass to the Lockheed Blackbird that can fly at 2,000 mph at 85,000 feet – what gets a plane into the air and keeps it there has not changed, except perhaps in the case of the Harrier jumpjet which rises vertically on sheer engine thrust before switching to normal flight.
To fly, wings that are moving horizontally through the air need to generate a powerful lifting force from the air moving past them. To do this, the upper surface of an aircraft wing is more rounded than the lower and the front of the wing more rounded than the rear. So a section through the wing, known as the aerofoil, looks like an elongated teardrop (fig. 7.1). The airflow, hitting the front edge, splits to pass over and under the wing, and produces lift. We all know that it works because we trust our lives with it but the reason it does so is somewhat counter-intuitive.
If one imagines the air seeded with visible particles, then the particles are carried by the air and their pathways are known as streamlines. Before the air hits the wing the flow is symmetrical, so the streamlines are parallel and equally spaced. When the air hits the wing, the airflow splits asymmetrically due to the curve of the wing: over the wing the particles, and hence their streamlines, are forced closer together. This means that the mass of air over the wing has to be squeezed through a tighter area than the corresponding mass below. The only way for the air over the wing to do that without compression is by moving faster than the air below. The streamlines are no longer parallel, but are curved and stay closer together over the wing (ie, the particles are carried faster above than below). The energy for this extra speed comes from the air pressure, which falls as a result. Thus the pressure above the wing is lower than the pressure below and the result is an upward pressure on the wing: lift.
There are fundamental differences between aeroplane and insect flight. For a heavy aeroplane, the main battle is with gravity. But for small creatures, gravity is much less important – insects can fall from a great height and come to no harm. This is because air resistance depends on the surface area presented by a body whereas gravity acts on the three-dimensional mass. As we saw with the gecko, as bodies get bigger their mass becomes disproportionately large compared to their surface area, and vice versa. So small creatures fall through the air slowly: they are, in effect, their own parachutes. All things left to fall through the air attain a terminal velocity at the point at which the air resistance matches gravity: a sphere of 1 m diameter reaches a terminal velocity in air of 1,138 kph, one of 1 mm only 13 kph. But when it comes to flapping flight, if the air breaks the fall of a light creature it will also restrict its flapping motion to some degree.
With insect wings we are in new territory compared to the structures that give us the Lotus-Effect, the gecko’s adhesion and spider silk: insect wings are moving parts and they move fast. It is often said that nature and the human engineer have completely different approaches to movement. We use wheels, axles, ball bearings and linkages whilst nature uses stretchy muscles and bendy hinges. A fly’s wings beat at up to 200 times a second but they have no bearings. That ferocious vibration is sustained by a hinge made of resilin – this is the resilient material. So for the fly, nature had to come up with a hinge that can flex millions of times without breaking. Human engineers find this challenging.
The patterns made in the air by an insect’s wings are complex and its degree of control very subtle, but the actual wing movements are restricted. There are three basic movements and these can be combined in varying degrees. Simple flapping is just mechanically up and down with the wing held horizontal. This would not produce much lift or forward thrust because the forces on the upstrokes and downstrokes more or less cancel each other out. Sweeping is moving the wings backwards and forwards in the plane of the insect. Again, sweeping with no other movement would be neutral in terms of moving the insect through the air. The third motion is to twist the wing through an angle.
In a typical wingbeat cycle these movements are combined. The wings sweep forward and plunge downwards, tilted down at an angle of about 30-45°, and at the bottom of the stroke they reverse through up to 180° and sweep back again, turning over 180° again at the top (fig. 7.2). The wingtip traces out an approximate figure-of-eight pattern. It is the asymmetry of the downstroke and upstroke that creates the lift and thrust just as a child on a swing moves the body to give propulsion.
The upstroke either achieves some lift (as in the fly) or presents minimum drag (the dragonfly). The wings themselves are not rigid although they do not have any control muscles beyond the root. This means that any flexing of the wing has to be accomplished by a combination of movements by the root muscles and the ability of parts of the wing to ‘give’ under pressure from the air and thus change shape. Some insects, for example the blowfly, have a subtle additional control mechanism, a miniature gearbox that selects different degrees of leverage for different kinds of flight.
The problem in modelling insect flight is simply stated: in insects the wings move, in planes they stay still. You can put a model plane in a wind tunnel and, using smoke, see the regular patterns of airflow over the wings. The pressures and forces generated can be measured, and the equations of flight calculated. But when a wing flaps, the conditions around the wing change at each stage of the stroke, and the patterns made in the air currents are very complicated.
But, as with the Lotus-Effect and the scanning electron microscope, modern instrumentation, in this case high-speed video, came to the rescue. In 1996, Charlie Ellington, at Cambridge University, capped some 20 years of research into insect flight with a breakthrough study of tethered hawkmoths and a large-scale mechanical flapper that mimicked the hawkmoth’s behaviour.
Ellington is a quietly spoken American who came to Cambridge in 1973, intending to continue the studies he had begun on the swimming of fishes, but he was assigned insect flight instead. His office is a museum of biomechanics, because besides Ellington’s flapper, now pensioned off, it contains boxes of equipment from an illustrious predecessor, Professor James Gray, who published notable early studies on animal locomotion in the 1930s, especially on the swimming of dolphins.
Ellington published strong papers for two decades which defined the problem of insect flight rather than solved it. In 1984, it was he who showed that standard aeroplane theory could not account for the lift generated by insects, but at the time he was unable to identify the actual cause. Photographs of smoke patterns in wind tunnels looked suggestive but were difficult to interpret because the air patterns generated by the wings are three-dimensional. But all the time the instrumentation was improving.
In 1996, using stereophotography to capture the flows of smoke around his tethered hawkmoths, Ellington saw vortexes – whirls of air like miniature tornadoes – form along the front of the wing on the downstroke to create additional lift (fig. 7.3). To be able to control the experiment better (butterflies and moths are not well behaved in the laboratory) he then built a mechanical flapper about 10 times bigger than the hawkmoth. To preserve aerodynamic similarity, he had the flapper beat its wings at 0.3 beats per second whereas the hawkmoth beats 26 times a second. Models on a larger scale than the actual insects are often used in this work. Michael Dickinson, at Caltech, the leading US expert in insect flight, uses a model fruitfly* 100 times life size. To duplicate the fly’s aerodynamics on this scale requires the adjustment of several conditions. There is a connection between the size and frequency of flapping of a wing and the viscosity (thickness) of the medium. This means that an insect wing can be modelled by a much larger wing that flaps more slowly in a much more viscous medium, in Michael Dickinson’s case, oil.
Although Ellington’s work on the flight of real insects was a great inspiration to the MAV movement, he is in the tradition of biomechanic scepticism. One of his first remarks when I met him was: ‘We’ve had so much hype with MAVs – it was ridiculous what happened with them. I lost count but the last time I added up they’d spent $55 million on micro air vehicles and they would have done better to have given $10,000 to a couple of good aero-modellers.’ And he groaned. So, although engineers and biologists work together on insect flight, they often see it from very different angles.
Ron Fearing says of his biologist colleague Michael Dickinson: ‘The idea that we could make a flying robot based on his principles is amusing to him and maybe at some point it would provide him with some results that would be important for the things he’s looking at, but it’s not the thing of real importance, which is understanding the secrets of nature. Michael Dickinson’s a great engineer too, when you look at all the apparatus he’s had to design just to do his experiments, to get his data, but the engineering is incidental to unlocking the secrets.’
Ellington’s work stimulated other laboratories to look at the question and for a while, as so often happens after an initial breakthrough, the position seemed to become more confused. The fact that everyone is studying different insects does not help matters and it is clear that there is no one answer to the question: ‘How do insects fly?’ In reality, they are versatile and use several different mechanisms. Studies on free-flying red admiral butterflies, a few years after Ellington’s work, concluded: ‘The micro-air vehicle community may find it daunting that the first flow visualizations of free-flying insects have revealed such a wide range of aerodynamic mechanisms, and that the insect switches between them on successive wingstrokes with such apparent ease.’
But Ron Fearing was certainly not daunted. He has a practical goal and this helps him to cut through some of the confusion. Concerning the various mechanisms he says: ‘I think our range of aerodynamic mechanisms is simpler than in butterfly wings. Butterfly wings are so large. You’re worried about the compliance [bendiness] of the wings. Our model is the house fly.’
But, given the complexities of flapping flight, is it necessary for an MAV to flap its wings? Fixed-wing planes are not very manoeuvrable and they cannot hover – both highly desirable traits in a small surveillance craft – but when Darpa held a competition in 1999 the winner, the Black Widow from the Californian AeroVironment Inc., was a fixed-wing plane. This led to jokes about Darpa spending millions on feeding the hobbies of overgrown schoolboy aero-modellers (fig 7.4).
What about a micro-helicopter? Both Fearing and Żbikowski reject this and their reasons reveal their biases. Ron Fearing says: ‘When you get very small, the ball-bearings in a helicopter rotor are a problem. I wouldn’t bet someone a lot of money that they couldn’t do it – make a small helicopter – but the fly is the favourite.’ For Rafal Żbikowski the problems are that helicopters lose lift whenever they come close to a wall (you don’t fly helicopters into the Grand Canyon); they are also very noisy and have high fuel consumption.
In 1996, Rafal Żbikowski was a young control engineer at Glasgow University when he heard of Darpa’s project. He liked the sound of it except for one aspect: Darpa specified both indoor and outdoor use but Żbikowski felt that the gap was in indoor surveillance, both military (think of the cave complexes used by terrorist organizations such as Al Qaeda) and civilian – the D3 jobs (dull, dirty and dangerous). Rescuing people trapped under rubble is the classic mission. Zbikowski says: ‘Now they use sniffer dogs, listening devices and thermal cameras. They have to be very careful how they lift the slabs of concrete to avoid a local avalanche. But if there is an opening and someone is breathing, they can fly a small thing in there, have a good look at it and then plan the operation better.’
Żbikowski now works at the Cranfield University campus at the Royal Military College of Science, Shrivenham. Cranfield operates both as an academic institution doing open research and also as a contractor to the UK Ministry of Defence (MoD), giving technological advice and support to the military. Żbikowski is Polish, with a wry sense of humour and a bustling organized manner. Aeronautics on this scale is highly technical and mathematical but when he gives talks he says pronouncing his name is the only hard part.
To him, it is the need for rapid manoeuvrability in confined spaces that tips the balance in favour of flapping flight: ‘It’s not that I woke up one morning and said, hey, let’s do something crazy: flapping like insects. The question is: what is the envelope for indoor flight; how can we realize it with a proven technology, something we know in advance is already working. This is how we reached this conclusion; you just look in your garden and it works.’ (Echoes of Feynman there, but this time you don’t need a microscope.)
Thinking about insect flight, you really do have to look in the garden from time to time, pinch yourself and say: ‘This isn’t just a lazy summer’s day with bees, hoverflies and butterflies as bit-players around the flowers: this is micro-aerobatics in action.’
Żbikowski’s and Fearing’s approaches are similar in everything except scale. They are both ambitious programmes and many problems need to be solved. An MAV must be able to sense and avoid any obstacles, take photographs, record data, and find its way back home. To build an MAV you need a tiny power source, a wing-actuating system and a control mechanism.
Ron Fearing began work on the MAV – his version is known as the Micromechanical Flying Insect (MFI) – in 1998 with grants from Darpa and ONR MURI Biomimetic Robotics. His team has taken the approach that since a fly or an MAV has many systems, all of which are essential to achieve autonomous flight, the key systems need to be investigated simultaneously. Since 1998, wing mechanisms, sensor systems and control systems have all made good progress. The wing mechanism is an ingenious contrivance that, if it has not yet matched the fly, comes close in the essentials.
The model is the blowfly, which has a wing span of 25 mm and flaps 150 times a second. So it is quite big as flies go and the high wing speed is the reason that you can hear a fly buzzing but not a butterfly with its more sedate flapping. How do you drive a wing that fast and make it do all those fancy rotations? The best way is to make it resonate.
Resonance is another word for vibration – it simply means something that moves backwards and forwards in a regular rhythm. But it also has a metaphoric meaning – we often say that one thing resonates with another or that a particular experience is ‘resonant’: this gives us a clue to the deeper meaning of resonance. The idea of resonating with something or someone implies sympathy and on a technical level, as well as emotionally, this means that in order to resonate two things have to be ‘in tune’.
You can see how this works with the ‘whistling wineglass’ experiment. If you run a wetted finger around the rim of a large champagne glass (not a flute), at a certain speed it will begin to whine. It always makes the same note and if you rub faster or slower the note grows louder then dies away. There is just one resonant speed for a particular wineglass and one characteristic note that it emits.
Every structure has a natural resonant frequency – this was the problem with the Millennium Bridge over the Thames – the Wobbly Bridge. Before being modified, if large numbers of people walked in step over the bridge, it settled into resonance at the same frequency as the footsteps – this generated large oscillations, and, if left unchecked, the bridge could have progressively shaken itself to pieces. This fate actually befell the Tacoma Narrows bridge, across Puget Sound, Washington. This was the slimmest most graceful suspension bridge yet built, but its deck resonated in light winds; on 7 November 1940, only four months after it opened, it shook itself to pieces in a wind of only 70 kph.
Not everything resonates. For instance, you can’t in any meaningful way make a supertanker resonate like this. But there is another important difference between the supertanker and things that do resonate. When a tanker comes to a stop there is no energy stored to begin the reverse direction: it stops dead. In resonance, the end of each movement or stroke builds up energy to kickstart the subsequent movement or backstroke, and so it continues.
A yo-yo and a Slinky® are good examples of this. The weight of the yo-yo coming down stretches the string and this energy pulls the yo-yo back when it reaches the bottom (the jerk of hitting the bottom also automatically reverses the rotation). Coming up, the blow when it hits the hand reverses the rotation again and the forces of the blow and gravity send it unravelling down. A Slinky is a spring and it is a combination of gravity and the energy in the spring that keeps it flipping over.
Now it would be possible to devise a mechanism for a fly’s wing that operated on the supertanker principle. If energy had to be supplied to bring it to a halt at the end of each stroke and then to start it off again in the reverse direction, a fly would only manage a few strokes before it was exhausted. A real fly operates on the yo-yo principle and so must the artificial fly. Ron Fearing says: ‘If it’s not running at resonance it means that the flight muscle is doing all this work accelerating and decelerating the wing. The power should be going into accelerating and decelerating the air.’
When I met him at Berkeley in February 2004, Fearing showed me his latest version of the MFI (fig. 7.5). What was fascinating was that there was no attempt to copy the fly’s materials. Fearing stresses that modern engineering materials such as carbon fibre often have properties in advance of nature’s: ‘carbon fibre beats chitin,’ he says. The wing-flexing mechanism, which was originally steel, is now cut and folded from a sheet of carbon fibre – one of the uses of origami discussed in Chapter 8. The latest version has a seriously thin polyester wing with carbon-fibre reinforcing ribs. Fearing admits that ‘the insect wing hinge is very sophisticated – I don’t think we’ve beaten that. But strength to weight the MFI probably beats the fly. The MFI looks big but it weighs less than a fly – the fly’s full of water.’
A wing mechanism that can execute the basic figure-of-eight wingbeat cycle is the bedrock but by itself such a mechanism is unstable – the insect has complex control processes to maintain any desired orientation, whether hovering or forward flight or turning, diving or climbing. Planes change direction gradually, partly because they contain at least one person and abrupt turns put a terrible strain on the body: turns are measured in G-force.*
There is a Russian plane – the Su 27 – that amazes airshow audiences the world over with its ability seemingly to stop in the air, and sit on its tail in a blur of shuddering metal and shimmering heat haze, a manoeuvre known as the Pugachev cobra after the Russian test pilot Viktor Pugachev who first performed the manoeuvre in 1989. For a fly, this is routine. Ron Fearing says: ‘Michael Dickinson looked at a fly flying in a straight line and making 90° turns. That’s about as fast a turn as you can imagine. In less than 10 wingbeats there’s a 90° turn. That’s one 1/20 of a second to make a complete turn. You’d think there would be something dramatic to account for the right-angled turn, but at first they didn’t see anything different between the straight-line flight and the right-angled turn. So they did a very careful sifting of the data and they saw: oh, there’s a 1% variation in the wing stroke amplitude. At the moment, building the left and right wings so that they’re within 10% of each other is actually pretty good! And now to control it we know we probably need to be within 0.1% if 1% is all it takes to do this dramatic right-angled turn. If you get this wrong, in five wingbeats you can be completely on your back, in an unrecoverable dive. It’s a very unstable craft.’
Intriguingly, the ‘instability’ of insects brings them close to the current technology of high-performance jet fighters such as the Eurofighter and the Lockheed F22 Raptor. In flight, stability and agility are always at war. Insects are designed for agility and planes for stability. It is the need for stability that leads to tailplanes on aircraft: insects never have anything like the tailplanes of aircraft. By definition, stable means resistant to change but fighters need to change their flight pattern in a hurry – that is the essence of being a fighter.
Modern jet fighters can be unstable because, like the insects, they have automatic mechanisms to monitor constantly for instability and to correct it. A pilot would not be able to react quickly enough to do this. So the Eurofighter and the F22 have deliberately unstable flight dynamics, controlled by sophisticated flight-control systems, that enable them to be more agile in aerobatics than previous generations of fighters. It might seem perverse – not to say dangerous – to make a plane deliberately unstable, but it works.
Everyone stresses that insects don’t fly by ‘deciding’ how to move every part of the wing. There are no muscles beyond the roots of the wing so wing motion is dictated by the force and direction of these muscles, the bendiness of different parts of the wing and the force of the air acting on it. The fabric of the wings is so made that they adjust their shape in ways favourable to the motion under way. A simple example: if a sheet of material is curved like an aircraft’s aerofoil, it changes the way it flexes. You can model this in paper: take a sheet of A4, bend it slightly in the lengthwise direction as if you’re going to roll it up into a cylinder. If you grip one end and flap it, the sheet stretches taut on the downstroke and bends on the up. This is far too simple as a working wing but it does behave very differently on the up and downstrokes, which is the first thing that you need, pushing hard against the air on the down, taking the line of least resistance on the up. If wings did the same things in both directions then the forces would cancel out and there would be no lift.
One big question for MAV developers is: how much of the insect wing do they have to copy? The hinge where the wing joins the body is relatively long in most insects, restricting the amount of bending; by making the wing bendy enough to twist in the upstroke, despite the rigidity of the base, the wing can achieve the necessary profile on the upstroke. In any case, if the fly is the model, with its very small wing, too much twisting is not desirable. As Ron Fearing says, if the wing were too bendy ‘it would be like paddling a canoe with a cardboard paddle’.
At the 15 cm scale Rafal Żbikowski is working at, a flexible wing is necessary and his team have shown that quite simple synthetic wings, with only two control ‘muscles’ at the root, can bend in ways convincingly like the real thing. Żbikowski has a test bed model of his wing system, in which an ingenious linkage mechanism generates the figure-of-eight pattern of flapping flight (fig. 7.6).
For both Fearing and 
bikowski, the wing-flexing system, although complex, seems a manageable problem. The big question is control. The MAV is a self-steering robot, which means that it has to sense its environment and give appropriate control signals to the wing flexors. The more we know about insects’ ability in this department the more impressive they seem.
The fly’s sensory system is highly developed. Under the microscope, the head of a fly looks like a highly tooled piece of engineering (fig. 7.7). In their large compound eyes, the smaller species have several hundred and the larger ones several thousand individual eyes, each one of which views a tiny section of the visual field. The eyes give almost 360° vision and there are additional aids. There are three light-sensitive cells called ocelli on the top of the head which give the fly a constant sense of up and down, so that it does not become lost in a tumbling 360° maze. Antennae and wind-sensitive hairs also abound.
The fly’s neatest control mechanism is probably the halteres. These are a development in the larger, more advanced flies (the ones we are most familiar with) of the hind wings, which have become adapted as a gyroscope.* Staying upright is a major problem for an unstable flying platform with wings beating 150 times a second. The halteres beat out of phase with the wings and resemble pendulums, giving the fly a constant reference point for its motion in three dimensions. In the engineered ‘fly’, the halteres can be mimicked by miniature gyroscopes, with photocells serving as ocelli.
Each one of the fly’s hundreds of compound eye elements experiences a directional flow of sensations during motion – the scientists call this optic flow. The nervous system and brain need to analyse this data to deduce the fly’s position and direction of motion relative to its surroundings. Flies can process a prodigious amount of data: 17-18 pictures a second give us the appearance of continuous motion; the fly needs 150-200 pictures per second, the same speed as its wingbeat. This is one reason why flies are so hard to catch: they perceive and act upon the world far quicker than we do.
The basic theory of this kind of motion detection was worked out as long ago as the early 1950s. And there are computer simulations of how the fly uses the optic flow to correct deviations from a straight-line course. This is one important aspect of a fly’s repertoire – others, such as landing, are more complicated.
Rafał Żbikowski explains that a major difference between insects and aeroplanes is that aeroplanes have few sensors whereas insects have hundreds of them. The central control for an insect’s flight is relatively crude (this is where the less computing power than a toaster comes in), so it relies on local feedback loops between these many sensors and the actuators. A modern aircraft such as the Eurofighter relies on only some 20 different measurements feeding in to its control system, compared to 80,000 for the fly. If you have 20 measurements you can only have 20 different independent feedback loops, and the plane’s on-board computer has to solve some very complex equations. But the insect has so many feedback loops it does not need to solve these equations. Żbikowski stresses that if you can measure everything you don’t need to compute so much, and this seems to be the way that insects do it. It is a little bit like the way we execute a skill such as throwing a ball. We don’t have to solve equations to do it, the way that digital computers do: we look at where we want to send the ball and the feedback systems between our nerves and muscles, based on past experience, do the rest. Żbikowski believes that this sensor-rich feedback control system will be the key to controlling MAVs.
So how long will it take for all the components to come together and an MAV take to the air to execute a few manoeuvres under its own power?* Rafał Żbikowski says: ‘In five years we will have something that is mechanically viable as a flapping device, ie, a reference platform that will have a chance of getting airborne.’ Ron Fearing’s mechanisms have operated statically in wind tunnels, and one has produced motion when attached to an arm like a record player stylus. As Fearing says: ‘You could make an MAV yourself. You can buy a rubber band powered flapper for $10.’ But a rubber band only lasts seconds: they are aiming for the real thing.
Ron Fearing believes that this is not too far away: ‘We’re starting to find that robots can be smarter than insects – in the eighties it was clear that the insects were more capable. I hope in this decade we’ll see robots being more capable than insects.’ The current target for an MFI flight is five minutes, limited by the battery. But Fearing points out: ‘5 minutes = 300 seconds, at 3 m/s gives a kilometre range. In a building, or an urban environment, you can’t go that far without being out of radio range. If a building is on fire, and you can’t find victims in 5 minutes, the fly would not be that useful. Having it last one hour in this scenario would not greatly increase its utility. Alternating between perching 95% of the time and flying 5% is a good strategy for longer duration missions. Perching doesn’t take much battery power.’
The MFI is the closest thing on the stocks to those swarms of nanobots Michael Crichton let loose on our imaginations in Prey. Indeed, Fearing does talk about releasing them in swarms, the point being that each individual MFI is expendable and it wouldn’t matter if it did not complete its mission. Only one needs to get through. But if an MFI gets out of control it is not going to unleash the mayhem seen in Prey: it will merely flop onto the ground, spent. The MFI is a small machine, made from carbon fibre and polyester, some steel and other bits and pieces. There is no way such a creature could be made from biological material in even the distant future. But now we can see what it really does take to make a small, self-controlling flying vehicle, we have to admire the originals – the flies, moths, butterflies and bees – all the more for it.