The buzz about insect robots
Key text
This topic is sponsored by the Sir Mark Oliphant International Frontiers of Science and Technology Conference Series.
Insects are a marvel of bioengineering. Scientists are studying how insects walk, fly and navigate their way in the world to overcome some of the obstacles in the development of robots.
How insects are inspiring a new wave of robotics
Insects are rarely held in high regard. Mosquitoes bite, flies annoy, locusts destroy crops, and cockroaches just give us the creeps. Perhaps that's why scientific research has most often focused on how to destroy them rather than on learning about the exquisite ways in which they function. How does a fly manouevre with such precision, a bee find its way from flower to hive, or a cockroach move so quickly?
Scientists have only recently started answering such questions. And increasingly they realise that insects are the superheroes of the planet, with unique attributes that could guide the next wave of advances in biomimetic robotics.
What is biomimetics?
Biomimetics is a term used for those engineering systems that make use of traits observed in biology. It may be a relatively new term, but the practice of borrowing from nature is old. Even the earliest stone-age implements mimicked attributes of other animals, such as claws used for digging or teeth used for ripping meat.
Related site: Asimo, the Honda humanoid robot
Asimo is the world's most advanced humanoid robot designed by Honda researchers to fit in with the office or home environment.
(Honda Worldwide)
We have grown accustomed to biomimetics in robotics, at least in science fiction. From the vain but decent C3PO in Star Wars to the ever-complaining Marvin in Hitchhiker's Guide to the Galaxy, the robots we've come to know and love have often exhibited distinctly human traits, both physical and psychological. Mostly, we want robots to perform tasks that humans would rather not do – or can't do – themselves, such as clearing land-mines or exploring hostile planets. This has meant that we've tended to visualise robots as humanoid, often with arms and legs and the capacity to solve problems.
But problem-solving – the ability to assess the elements of a problem and to generate and evaluate possible solutions – is a very difficult skill to install in a machine. Even now, after decades of development, a 'thinking' robot may struggle with the task of walking around the block. This limitation has held back the development of robots that can truly perform some of the less pleasant or more dangerous tasks that humans are currently required to do. Moreover, the usual means of robot locomotion – on wheels or simple sets of legs – are fine when tested on a smooth laboratory floor but generally inadequate on uneven terrain.
What's so special about insects?
Entomologists have often marvelled at and puzzled over the difficult tasks that insects can perform, despite the apparent absence of a great deal of brainpower. We all know how successful they are: there are probably more than four million species of insect on the planet compared to about 4500 species of mammal. Perhaps by learning about and then mimicking the attributes of insects, which have been honed by millions of years of evolution, we can fast-track the development of robots – so that they can do what we want them to do without needing brains the size of a planet.
Insects have some practical advantages, too. They are plentiful and easy to maintain in laboratories and they have an exoskeleton (a hard outer covering) rather than an internal skeleton like humans, making the study of the way they move relatively easy. They are also immensely diverse, offering a wide variety of strategies for things like locomotion, navigation and vision. And recent advances in miniaturisation – the development of technologies in increasingly small sizes – make the construction of insect-sized robots possible.
The masterful scuttle of the cockroach
Of all the insect superheroes, the cockroach is perhaps the most fabulous of all. One species has been clocked at speeds of up to 50 body-lengths per second (few humans can run faster than five or six) and another species has been observed to scuttle across uneven terrain and surmount obstacles higher than itself without slowing down. Scientists have developed six-legged robots before – largely because of the stability they offer – but have been unable to achieve much in the way of speed.
Cockroaches are showing us how it can be done. The cockroach is shaped the way it is to maximise its speed and stability and to enable it to squeeze into narrow cracks. Detailed studies of their locomotion have also revealed several design principles that can be used in biomimetic robots. These include a self-stabilising posture – achieved through a low centre of mass located towards the rear of the animal and by a wide base of support – and a thrusting leg function, in which the legs act mainly as thrusters rather than striders, launching the insect forward.
Robotics engineers are now hard at work designing machines that mimic, as much as possible, the biomechanics of the cockroach. The idea is that with a properly designed system there will be much less need for computational ability – no longer should robots have to spend time calculating how to take their next step to ensure they don't fall over; it should be instinctive. Scientists at Stanford University in the United States have created what they call the iSprawl robot, which adapts some of the principles of cockroach locomotion. One version of iSprawl is about 11 centimetres long and can move at about 15 body-lengths per second.
Making bee-lines and nosing around
Another difficult problem for robots is navigation. How can they find their way somewhere and then back again without spending all their time (and computer power) thinking about it? Again, insects might have the answer. A bee, for example, astounds scientists by wandering in an apparently aimless way to a nectar source and then returning to the hive sometimes several kilometres away – all with a brain that contains fewer than a million neurons. A supercomputer (and even a human) might struggle to accomplish such a task.
Australian scientists are among those who have conducted detailed studies of bee navigation. They found that bees use several strategies for navigating their flower-strewn way. For example, they use the sun as a compass to determine flight direction, even if it's behind a cloud, and they can allow for the tracking of the sun across the sky using an internal 'clock'. Bees can also store simple information about the places they've been and recall this on their return flight, making connections between landmarks and their location with respect to the hive. And they use something called optic flow to judge their flight distance and to negotiate confined spaces.
These strategies are of immense interest to robotics engineers. Adapting them for use by robots could cut down significantly on expensive, power-consuming equipment such as geographic information systems and satellite navigation systems, which might not always be available.
Flight of the bumble bee
We all know that insects are brilliant aviators: it's probably no coincidence that one insect, the fly, is so named for its extraordinary ability in the air. But an engineer once calculated, infamously, that an insect (in this case, a bumblebee) couldn't fly, at least not by the steady-state principles of aeronautics that he applied. An aeroplane with proportionally the same weight and wing size would certainly never get off the ground. The solution to the puzzle of insect flight had to wait for technological advances – such as high-resolution, fast-frame photography – that enabled the detailed study of an insect's tiny wings and their rapid movement through the air.
It turns out that insects use a combination of three aerodynamic techniques to fly and to perform their astounding aerobatics. One is called delayed stall. Insect wings flap at a very steep angle, which would lead to the stalling of flight, except that in flapping like this the wing generates what is called a leading edge vortex. This in turn produces an area of low pressure on the top surface of the wing, thereby pulling it upwards. A second technique is called rotational circulation. The insect rotates its wings at the end of each stroke, inducing an area of low pressure to generate additional lift. The third technique is to extract energy from the wake of the previous wing stroke.
These insights into insect flight could be used in robotics in a couple of different ways. They could help in the design of small, flighted robots that could perform useful tasks in circumstances where a large payload was not necessary: search-and-rescue, for example. And they could be used in the exploration of other planets, where a thicker atmosphere would enable larger robots to use the same flight strategies. The US Department of Defense, which is investing in the development of these small flying robots, has dubbed them 'micro air vehicles' (MAVs). It sees uses for them in surveillance and reconnaissance. Indeed, a large part of the funding for this kind of work comes from defence organisations, so it's likely that many of the initial applications will be military in nature – civilian applications may come later.
So little brain, so much skill
The adaptation of insect survival strategies to robotics is still in its early stages, and there are few functioning examples. Nevertheless, the possibilities are almost unlimited. The processes of evolution have been honing nature's gifts for millions of insect species for millennia, and it would indeed be surprising if we could learn nothing from them. Many of the solutions to complex problems that nature has come up with are magnificent in their simplicity, yet adapting even these to our purposes will not be easy. It will require a great deal of brainpower – from the robotics scientists that are setting out to do so. It seems likely, though, that robotic superheroes really will cruise the planet some day, even if sometimes we hardly know they're there.
Posted May 2005.