麻豆传媒

IN COUNTLESS B-movies giant alien insectoids invade the Earth and wreak havoc, trampling through cities and tearing down buildings. Now we puny earthlings are hoping to turn the tables, and send insect-like robots to investigate the surface of Mars.

These 鈥渂iomimetic鈥� robots that can walk, climb and fly like real creatures are already under development 鈥� and some can almost think like them. Such designs promise to be more agile, robust and productive than the wheeled vehicles such as NASA鈥檚 Mars rovers Spirit and Opportunity. Wheeled robots only work when the ground is flat and firm, with few boulders. Opportunity has only just freed itself after a month stuck on a 30-centimetre-high sand dune. Giving robots the ability to walk like insects, on jointed legs, makes them far more agile, allowing them to cope with large obstacles, sandy surfaces and steep inclines.

At the Biologically Inspired Robotics Laboratory at Case Western Reserve University in Cleveland, Ohio, Roger Quinn and his team have built a six-legged walker inspired by the death鈥檚 head cockroach. Called Robot V, it is 20 times the size of the real-life insect and was designed to allow the study of legged locomotion over uneven terrain. But Robot V, like most walker designs, has a major snag: it is extremely complicated. Its legs have up to five segments, each one operated by its own artificial pneumatic muscle driven by compressed air. This means it consumes a lot of power and has a large number of components that could fail.

鈥淕iving robots the ability to walk like insects, on jointed legs, makes them far more agile, allowing them to cope with large obstacles鈥�

So Quinn has developed a concept that combines the robustness of wheels with the agility of legs. The result is Whegs II, a robot with six wheel-leg hybrids, or 鈥渨hegs鈥�. Each wheg consists of a central rotating hub attached to an axle, from which three flexible legs protrude, each with its own gripped foot. As the axle turns the hub, the legs spin round and make contact with the ground. This arrangement allows a single motor to drive all six whegs. 鈥淭his keeps its weight and complexity down,鈥� Quinn says. He also says the design would be easy to scale up.

The robot can run fast and climb over obstacles like a cockroach. Two forward-facing antennae also allow it to sense whether an object is best avoided, climbed over or tunnelled under. To overcome larger obstacles, Whegs II flexes in the middle, allowing the thorax to rear up and the front wheel-legs to pull it over the top 鈥� just as a cockroach does.

Conventional robotic rovers have other drawbacks too. Spirit and Opportunity, for example, are remotely controlled from Earth, which entails a frustrating wait of up to 40 minutes after each command is sent. The hope is therefore to design autonomous explorers that can do some thinking for themselves, and so manage without 24-hour babysitting. With complex six-legged walkers, the task of programming the precisely timed pattern of leg movements is impossibly complicated. So researchers have been turning to biologically inspired control methods.

Scorpion is an eight-legged walker being developed by Frank Kirchner and his team in the robotics group at the University of Bremen in Germany. In animals, groups of neurons called central pattern generators control the rhythmic contraction and relaxation of the muscles that perform actions such as walking. Scorpion has electronic circuits that mimic these neurons to produce a rhythmic signal that controls the motors operating each joint in the leg.

Scorpion is also equipped with sensors to detect the tilt of its body, the position of every joint and the load on each foot. This information is fed back into the circuits to keep the robot walking smoothly over uneven terrain. The software adapts to keep the gait steady over rocks, fine sand or steep inclines. If its sensors feel Scorpion starting to stumble they rapidly trigger a pre-programmed 鈥渞eflex action鈥� to stabilise it, says Dirk Spenneberg, senior scientist on the project. 鈥淲e have studied the control systems of a wide range of animals, from cats to stick insects to lobsters, in order to find the most interesting biological mechanisms to use for robot locomotion.鈥�

One thing Scorpion can cope with better than real animals is turning turtle: when it senses it has fallen over onto its back, the legs invert themselves, allowing it to carry on walking upside-down.

Despite the ingenuity of Robot V, Whegs, Scorpion and their kind, not a single biomimetic design has yet been adopted by NASA. Spacecraft engineers are a conservative bunch, very cautious about new technologies. While biomimetic strategies are in principle more versatile and robust than conventional designs, engineers argue that they have not yet been sufficiently refined or exhaustively tested. Standard wheeled rovers controlled from Earth perform adequately, they say, and have proved themselves reasonably reliable 鈥� so why risk a novel technology on a multimillion-dollar planetary mission?

Mars exploration programmes are all focused on plans that can produce a vehicle within the next five years or so, says Anthony Colozza at the NASA Institute for Advanced Concepts, who leads a group of NASA researchers that examines the feasibility of proposed designs. Most of the biomimetic proposals lie further in the future, he says. 鈥淐onsequently many find it difficult to get funding to continue their development.鈥�

But Alex Ellery, a robotics researcher at the Surrey Space Centre, part of the University of Surrey in the UK, believes biomimicry will soon come into its own. 鈥淚n about 15 years鈥� time, the scientific mission demands, such as searching for evidence of life in rugged terrain, will have outstripped the capabilities of wheeled rovers,鈥� he says.

Ellery is working with Julian Vincent, director of the Centre for Biomimetic and Natural Technologies at the University of Bath, UK, on a biological solution to a problem that conventional technology just cannot solve. Because Mars lacks a thick atmosphere, its surface has been exposed to radiation from space for billions of years, which has rendered the topsoil highly oxidising. It seems impossible that any form of life could survive under these conditions, and even complex organic molecules 鈥� the markers of life that astrobiologists hope to find 鈥� would be rapidly degraded. Top of the wish list for life hunters, therefore, is to dig beneath this destructive upper layer and analyse samples of the soil beneath. As this oxidising layer is thought to be a few metres thick, a conventional drill would need to be at least that long and therefore too big to fit onto a lander.

Ellery will this month submit a potential solution to this problem in a proposal to the UK鈥檚 Particle Physics and Astronomy Research Council (PPARC), as part of ESA鈥檚 Aurora programme to explore the solar system. The idea is based on the egg-laying apparatus, or ovipositor, of the woodwasp. This takes the form of a sharp spike that the wasp uses to drill into wood so it can place its eggs in a protected and nutritious location. The spike is split down the middle, and the two halves slide back and forth past each other. Each half has a row of backward-pointing teeth running along it so that as one half pushes through the wood, the teeth on the other dig in and hold fast.

The big advantage of a Martian probe that copied this design is that no push would be needed from above: the bit would simply pull itself into the Martian soil. The drill could be a self-contained unit capable of burrowing deep underground, which could later retract its teeth, allowing it to be pulled back up the shaft to deliver the samples. 鈥淭his kind of biologically inspired innovation is exactly what is needed to keep space missions returning new data,鈥� Ellery says.