快猫短视频

THE SPHERE ploughs into the planet鈥檚 dusty surface just as the Sun rises.
With a muffled clunk, the metal casing opens and something furry rolls out,
unfolds and raises itself slowly onto four legs. Most of its body is covered
with spiky purple hairs but across its flank some of the hairs are orange and
arranged in an indistinct pattern. Only when the thing turns towards the rising
Sun does the pattern become clear. The orange hairs spell out four letters:
N-A-S-A. Chang Liu鈥檚 hairy space probe has made it to Mars.

For the moment, Liu, a microelectronics researcher at the University of
Illinois at Urbana-Champaign, can only dream: he has a few years to wait before
he can send his furry creations into space. But he is confident that the
artificial hair he is 鈥済rowing鈥� in the lab has a great future. 鈥淲e鈥檝e only been
going half a year and we鈥檝e already produced the first artificial hair cell
array,鈥� he enthuses. 鈥淭his is going to go a long way.鈥�

All the way to Mars, perhaps. NASA has realised that it has a lot to learn
from nature about dependable sensors. There is only so much room on the surface
of a robotic explorer: when you need an array of one kind of detector to keep it
upright, moving and stable, and another kind to monitor an unfamiliar
environment for wind strength, strange textures or large obstacles, something
has to give. Unless one type of sensor could do it all. Something, for example,
like a hair. Wrap spacecraft, planetary landers鈥攅ven astronauts鈥�
suits鈥攊n a coat of sensitive, artificial fur and you could negotiate the
complexities of deep space with complete confidence.

Hair clearly works in nature, and not just as a thermal insulator. It gets
everywhere, monitoring an enormous variety of things, including orientation,
balance, speed, sound and pressure. 鈥淎nimals take a single structure and simply
by varying some details, adapt it for sensing different kinds of things,鈥� says
Fred Delcomyn, an insect neurobiologist at Urbana-Champaign who works with Liu.
A cockroach, for example, uses hair to sense its proximity to nearby objects.
The insect also uses long, slender hairs to detect air movement caused, perhaps,
by the stealthy approach of a predator. Other, shorter hairs grow in clumps on
the joints between its body and its legs where they can detect the relative
position of each leg. Fish are just as sensitive. They use hairs as pressure
sensors or as accelerometers, allowing them to assess their movement through the
water.

Copy the way these creatures utilise hairs and you can create an array of
artificial hairs to sense almost anything. 鈥淭hat way you don鈥檛 have to do a
whole new development process for every single kind of sensor,鈥� Delcomyn
says.

You might imagine this would be an easy job. Natural sensory hair is simple
stuff: a flexible filament attached to a nerve cell that can tell how much the
filament is bent. So why haven鈥檛 researchers already built the stuff?

Try thinking of a cheap and easy way to mass-produce an array of thousands of
vertical strands. Each strand must be able to flex without damage, record the
exact amount and direction of deformation, and then bounce back to its original
position. And all the strands must be connected to a network of microprocessors.
鈥淭his is quite a significant challenge,鈥� says Liu.

But he believes he can rise to it, thanks largely to his skill at silicon
origami. In an attempt to develop three-dimensional structures for circuit board
components, Liu has taken research started by Kris Pister, an electrical
engineer at the University of California, Berkeley, and turned it into an art
form, creating tiny hinged, folding flaps that can be assembled into whatever
shape is needed.

His early attempts to make hair created something more akin to scales. He
begins with a thin, flexible layer of glass and polycrystalline silicon
deposited on a silicon substrate. Next he etches the polycrystalline layer using
hydrofluoric acid, cutting hinges, flaps and anchoring slots. The hinges and the
flaps must be pulled up, but each is only around 100 micrometres across and
Liu can鈥檛 flick each one up by hand. Instead, he coats the flat arrangement with
a layer of a magnetic nickel iron alloy called Permalloy. He then applies a
steadily increasing magnetic field to the substrate, and as if by magic the
flaps rise in exactly the right order for them to slot together, forming robust
3D structures about a tenth of a millimetre across
(see Diagram).

It might look like magic, but it鈥檚 actually down to precision machining and
the distribution of Permalloy. If you want one flap to rise before its
neighbour, for example, simply coat it with slightly more of the magnetic
material so that it pops up in a lower magnetic field.

Few other micro-assembly techniques are quite so elegant, or so well suited
to mass-production, says Liu. 鈥淥ur technique is high yield, highly predictable,
and doesn鈥檛 require any dedicated surface area for actuators.鈥� NASA and the
National Science Foundation are certainly impressed: they recently handed him
$600 000 of funding to turn his scales into something more like hair.

His latest filament is a spike etched from silicon nitride, 20 micrometres in
diameter and between 20 and 500 micro-metres long. It is hinged to a silicon
substrate by a thin piece of gold wire. Liu also deposits a layer of Permalloy
along the far end of the 鈥渉air鈥�.

Like his scales, the spikes are made by building up layers of silicon nitride
on a silicon substrate, etching, then adding a layer of Permalloy. A small
magnetic field is enough to make the hairs stand on end
(see Diagram).
Near the base of each hair is a polysilicon piezo-resistor which is
hooked up to an electric circuit. When the filament is bent, it exerts a tiny
force on the piezoresistor. This alters the current flowing through the
resistor, revealing the size of the force applied to the filament.

Liu has already made an array of 15 hairs measuring over a micrometre across,
but will soon be creating larger expanses of hairy skin. 鈥淲ith this fabrication
process and assembly technique we should be able to make a really large and
continuous sheet,鈥� he says.

Eventually he aims to make the hair filament out of a soft, flexible polymer
material such as polyxylylene, with a thin, metal core. 鈥淲e want the hair to be
as soft as possible, so that it can take a lot of beating from the environment,鈥�
he says.

The final step will be to wrap the finished hairy skin around a waiting
probe. All will be complete鈥攅xcept for one small thing. How will the hairy
skin tell the probe what it is feeling?

鈥淭his, I think, is going to be the greatest challenge,鈥� says Delcomyn.
鈥淏iologists still have a lot of questions about how it鈥檚 done in nature.鈥� Even
small insects can have thousands of different signals assaulting their nervous
system at any one time. That is a big signal-processing problem, and one that
Liu and Delcomyn will have to resolve if the hairy space probe is to be of any
use.

Fortunately, nature uses some clever tricks to simplify the task. Look
closely at a cockroach鈥檚 rear end鈥攊f you dare鈥攁nd you鈥檒l see the
hairs are arranged in specific places and patterns. Each responds to air
movement from a particular direction, and the hairs are arranged in an
asymmetric pattern to tell the creature which direction the air is coming from,
and thus the best escape route from an approaching predator. The neuron from
each hair connects to one of the insect鈥檚 ganglia鈥攖he terminals for nerve
inputs鈥攊n a way that mirrors the pattern of hair distribution. Most
importantly, the ganglia acts like a localised brain, speeding up the insect鈥檚
response to stimuli.

Hair cell neurons can also fire in ways that depend on the animal鈥檚 posture.
鈥淚f a leg, for example, is extended, then a particular stimulation applied to a
hair will have one kind of result,鈥� says Delcomyn. 鈥淚f the leg is flexed it will
have a different result. The position of the leg biases certain neurons within
the nervous system.鈥�

This filtering appears to simplify the signal processing, and Delcomyn
believes artificial hairs could, in principle, mimic this. There are other ways
to reduce the processing burden, too. 鈥淣ature has all sorts of tricks for doing
signal-processing mechanically or passively, so you don鈥檛 even need neurons for
a large part of the processing,鈥� says Ron Fearing, a robotics engineer at the
University of California, Berkeley. Fearing is developing touch sensors for
robots. The human fingertip, he says, is a prime example of using mechanical
properties to make the task of signal processing easier.

There are up to 200 nerve endings in every square centimetre of fingertip,
and the elastic properties of the skin ensure that even if none of the endings
is stimulated directly, nothing is missed. Touch a pin to the skin between these
endings, for example, and the skin deforms elastically around it, spreading the
stimulus to a number of nerve endings and alerting you to the pressure.

On the other hand, if there is too much information for the nerve endings to
process easily, for example from an intricately textured surface, skin鈥檚
elasticity allows it to act as a 鈥渓ow pass鈥� filter鈥攕moothing out the bumps
and making it easier for the brain to process textures. Liu hopes his artificial
hairy skin could filter information in a similar way.

Calculating wind speed or aerodynamic parameters from the deflection of the
hairs could be far harder since the patterns of bending will be more subtle. You
could cover the skin with microprocessors to improve computational power, says
Fearing, but the key to processing signals as efficiently as biological systems
depends on finding the right algorithms. Nature may have found brilliant short
cuts that do away with the need for sophisticated signal processing, he
says.

While Liu believes he can mimic the way that simple nervous systems work, he
concedes that there is simply not enough known about biological signal
processing to copy more complex systems. So with the help of Doug Jones and
Naresh Shanhbag of the University of Illinois鈥檚 electrical engineering
department, he is designing a signal processing system using a more conventional
approach.

It鈥檚 easy enough to make an electrical connection between an individual
strain sensor and a processor using standard techniques. But once Liu starts
trying to do that for every hair, things are going to get鈥攚ell, hairy.
鈥淚nitially, we鈥檒l have a wire from each sensor, but as the number of connections
increases, pretty soon it becomes physically impossible,鈥� says Liu.

To minimise the complexity, the team will tie several sensors into a cluster,
and connect clusters into groups. They will also distribute small processors
throughout the skin. 鈥淭he goal is to do localised processing before sending the
signal to a central processor,鈥� says Liu.

The result would be a multilayered system. For example, if the probe stubbed
its hairy toe on a rock, signals from the toe hair cells might go only as far as
the foot鈥檚 processor. This could contact the leg processor, which would decide
whether the urgency of the situation merited involving the central processor. If
it did, the central processor might concentrate resources on lifting that
leg.

Even if they turn out to be pretty dumb, hairy robots will be more sensitive
than any NASA probe. At their most basic, the hairs will function like cat鈥檚
whiskers, allowing vehicles docking with a space station to know when they are
getting too close, for example, or help a robot arm reach into a cargo bay
without bashing the wall. More sophisticated systems could create a whole new
array of touch and balance sensors for planetary landers and robot
explorers.

鈥淭here鈥檚 a plethora of uses for these things: where they go is only limited
by the imagination,鈥� says Roger Crouch, who coordinates the project at NASA鈥檚
life and microgravity division. Take astronauts, he suggests. On space walks
it鈥檚 easy to get disoriented or lose your sense of balance. 鈥淵ou don鈥檛 know
which way is up, or how to get back to where you came from,鈥� says Crouch, an
ex-astronaut himself. Sometimes astronauts moving around on the outside of
spacecraft leave markers, like Hansel and Gretel, in order to keep track of
where they鈥檝e been. With a space suit covered in artificial hairs providing
orientation and balance, such problems might disappear.

Yet the future could be stranger still. Back on Mars, the hairy space probe
turns a little further on its spindly legs and, from what appears to be its
head, it begins to uncurl a strange appendage. Moments later, lifted towards the
sky, is a huge robotic elephant鈥檚 trunk. It may not be as subtle as hair, but
limbs such as trunks and tentacles are versatile and adept tools for sensing,
sampling and manipulation (see 鈥淲hen Hairy met Nelly鈥�).

The Sun is climbing high now, and NASA鈥檚 latest explorer sets off on its
task. Combined with the probe鈥檚 bushy hair, the robotic trunk creates an eerie
sight on the dusty Martian plain. Millennia after their extinction on Earth,
NASA has put a woolly mammoth on Mars.

While Chang Liu develops his hairs, Ian Walker and Michael Hannon, robotics
engineers at Clemson University in South Carolina, are using NASA funding to
copy another idea from nature. Their robot trunk is a column of metal discs of
gradually decreasing diameter. Springs and wire tendons join each disc, and the
whole thing is controlled by an electric-powered servo system that can move the
trunk into highly contorted shapes. Eventually, a Martian explorer equipped with
a trunk could easily sweep aside large boulders, and scan high rock ledges or
narrow cracks for signs of life.