快猫短视频

ABOUT ten years ago, electrical engineer Thomas Furness had a gloriously
simple idea. Why clutter your home or office with all those cumbersome computer
monitors and TV screens, he asked himself, when you could paint the images
directly into your eyes instead? Best of all, they would be sharper and
brighter.

And what could this mean for the millions of people worldwide with corneas
damaged from birth, accident or just plain old age, he wondered. People doomed
to a world without work, let alone the comfort of a good book or TV soap? And
just suppose you could help surgeons perform complex operations by exploiting
the information from the real operating theatre and, say, a set of virtual
X-rays spread out in the personal Cinerama within the eyeball? Or, more scarily,
suppose fighter pilots could learn to outmanoeuvre the enemy in the increasingly
complex theatre of modern warfare by consulting data beamed straight into their
eyes? These were revolutionary thoughts. But Furness was no lone inventor or
idle dreamer. He was a practical man, and as director of the Human Interface
Technology Laboratory (HITL) at the University of Washington in Seattle, he was
well placed to bring together a team of researchers to work on the problem. In
1993, Furness鈥檚 baby was finally born: a bulky tangle of wires, lenses and
lasers that they called the Virtual Retinal Display.

Laser vision

Crudely, the VRD treats the retina as if it were a screen, with laser light
standing in for the light from a projector. Information from a computer is
relayed to the device via optical fibre, wires or a radio link, where it is used
to control a scanner made from two tiny mirrors. Guided by the scanner, the
laser then 鈥渄raws鈥� the images across the retina.

To move from such sketchy principles to a device that people would actually
want to use to improve their lives or work usually takes commercial nous. So
HITL has licensed the technology to a Seattle-based company called Microvision.
With the extra money and research support that this brought, the design of the
VRD has evolved quickly from the prototype鈥攁 bulky, bench-mounted
dinosaur鈥攖o the most recent model that is worn attached to a helmet.

Most of its components are attached to a short metal arm which flips down
over the right eye. The control electronics and tiny mirrors that steer the
laser beam are mounted on this arm. At its base, just a few centimetres in front
of the eye, is an outsize monocle that bounces the laser beam into the eye,
where the cornea focuses the light onto the retina.

While the prototype was large, eventually, Furness predicts, the VRD will
look like a large pair of glasses which will obviously tidy up home or office
nicely. More importantly, says HITL鈥檚 Erik Viirre, because the images are drawn
with the intense light of a laser beam, images seen on the VRD look better than
the original. 鈥淚t鈥檚 like having a bright computer screen floating in the air,
but it鈥檚 actually much sharper, and the contrast is better.鈥�

The VRD creates those images by stealing a trick or two from computing. What
we see on a computer monitor is a picture made up from a pattern of millions of
points of coloured light. These tiny points of light, pixels, are too small to
see individually, but since each is a different colour and intensity, they work
together to fool the eye into seeing a single coherent image. To copy a picture
from a computer screen onto the retina, the VRD simply calculates the intensity,
colour and exact coordinates of each pixel. Laser light then dots the retina
with the same array of pixels by switching the beam on and off millions of times
each second.

To put each pixel in its place, the laser beam scans across the retina in a
series of horizontal lines, each one slightly beneath the last, following the
same pattern that your eyes trace as they read the lines on this page. The beam
is moved about on the retina using a pair of mirrors: the first rotates in the
vertical plane to move the beam up and down and the second rotates in the
horizontal plane to move the beam from left to right
(see Diagram).

The mirrors scan the laser over the retina at high speed鈥攊f you could
read this magazine at the same rate, you would finish the whole issue in just a
few seconds. And to prevent the image flickering, each full
frame鈥攃urrently containing about 1000 lines made up of more than 1000
pixels each鈥攎ust be redrawn at least 60 times each second.

But even this scanning speed isn鈥檛 fast enough to generate an image that is
truly lifelike. To get anywhere near the eye鈥檚 resolution, Furness must put more
detail into the image by cramming 15 times as many pixels into the picture, yet
still redraw the frame 60 times each second. In short, he must supercharge his
scanner.

Researchers at HITL and Microvision realised that the best way forward was to
make the mirrors smaller. Tiny mirrors are lighter, they reasoned, and with less
momentum they can be moved backwards and forwards faster. They also realised
there was a need for a new kind of high-speed motor to drive these minute
mirrors.

Initially, they mounted a tiny mirror only 6 millimetres across on each end
of a miniature metal seesaw so that as the seesaw rocked back and forth, the
mirrors moved with it. Minute coils of wire connected into an electrical circuit
were arranged beneath each end of the seesaw. To rock the seesaw at high speed,
a magnetic field which changes polarity 15 000 times a second is passed through
the coils. This alternately attracts then repels the ends of the seesaw up and
down.

But with its complicated components, the scanner is fiddly and expensive to
make, and at two centimetres long, it鈥檚 relatively bulky and heavy. To shrink
the scanner further, engineers at Microvision turned to the tiny world of
microelectromechanical systems for help. By etching thin layers from a sliver of
silicon, they have built a scanner that weighs a mere 5 grams and measures less
than 1 square centimetre. The mirror, too, is much smaller at 1 millimetre
across, and it鈥檚 mounted on the end of a thin, flexible bar which is anchored to
the silicon.

鈥淏ut if you look at it through a magnifying glass, you won鈥檛 see anything
that looks like a motor,鈥� says John Lewis, director of research at Microvision.
鈥淭he mirror is moved by static electricity.鈥� Microvision鈥檚 team has turned the
mirror into one plate of a capacitor, with the other plate formed by a small
area of silicon beneath it. Put a rapidly varying voltage across the two plates,
and the mirror will first be attracted, then repelled. Unlike its larger
brother, this mirror can move up and down more than 30 000 times each second,
says Lewis. And because the new improved scanner is built from silicon, building
large numbers should become cheap and simple because the engineers can exploit
existing chip manufacturing techniques.

But there is one more vital component that must be miniaturised if the VRD is
to shrink to the size of a pair of glasses: the laser itself. At the moment, the
VRD uses three separate lasers鈥攇reen, red and blue, each about the size of
a loaf of bread鈥攖o make up a full-colour image. These are mounted on a
bench and their light is combined and beamed to the VRD through a thin optical
fibre. Ideally, says Viirre, the lasers would be made using semiconductor
technology Tiny red semiconductor lasers less than 1 centimetre across are
already cheap and reliable.

Unfortunately, it鈥檚 a different story with semiconductor lasers that give out
blue or green light. At the moment, they are expensive, and only work for a few
hundred hours before burning out. Getting all three colours from tiny
semiconductor chips is certainly possible, says Viirre, but it may take the
industry a couple of years. When they do arrive, Viirre will finally be able to
pack his three large lasers away and replace them with miniature lasers mounted
directly onto the VRD.

One of the other problems with the VRD only becomes apparent when you put it
on. Viirre likens it to looking through a pair of high-magnification
binoculars鈥攜ou must line your eyes up precisely with the beam or the image
disappears. Since we rarely fix our eyes on any single point for more than a few
seconds, using the VRD becomes difficult. So Microvision is looking at an
eye-tracking system that follows the movements of the pupil by monitoring
reflections from the cornea. The tracker calculates where the eye is looking and
moves the lasers around to compensate. But this system is expensive and
complex.

Ultra-light lens

A better solution may lie with a special kind of lens known as a holographic
optical element. An HOE is actually a type of diffraction grating made by
recording a hologram inside a thin layer of polymer. It works by converting a
single laser beam into a circular array of 15 bright spots. Place the HOE
between the scanning mirrors and the eye, and the array of beams that forms will
illuminate the region round your pupil. Move your eye slightly and one of the
beams will still strike the cornea and be focussed to form an image on the
retina.

HOEs have a big advantage over eye-tracking systems: because they are made
from a thin layer of polymer, they weigh next to nothing. 鈥淎ll of the action
takes place in a layer just a fraction of a millimetre thick,鈥� says Lewis.

Since 1997, Microvision has signed contracts with the US Air Force, Boeing
and the Swedish-based aerospace company Ericsson-Saab Avionics to develop the
VRD to train the pilots who will fly the next generation of fighter planes.
Instead of the now-familiar virtual flight simulators, the VRD projects
realistic combat scenes into the pilot鈥檚 eyes. So the VRD could end up in Saab鈥檚
fourth-generation Gripen fighter and the US Air Force鈥檚 Comanche helicopter,
which are both scheduled to enter service by 2006.

While Microvision woos the military, back at HITL, Viirre is working with a
small group of partially sighted people. Partial loss of sight affects almost
two per cent of the population in the US alone, about five million people.
Almost half are of working age, yet their poor eye sight prevents them earning a
living or even reading.

For those whose sight loss is caused by a damaged or scared cornea, Viirre
believes that the bright images generated by the VRD could make a real
difference. When the cornea is damaged, images that strike the retina are
blurred. But a laser can pass through a distorted lens with little difficulty.
Viirre鈥檚 optimism seems well founded. 鈥淪o far, we鈥檝e had some pretty remarkable
results,鈥� he says. 鈥淥ne woman said it was like having a new cornea.鈥� Viirre鈥檚
quick to stress that it鈥檚 not a cure-all, however. 鈥淚t won鈥檛 work with complete
loss of the retina, and it offers little help to those who have lost the central
portion of their vision.鈥�

And there is no danger that the laser could cause further eye damage,
Microvision claims. The power of the lasers in the VRD are just a few hundred
nanowatts: Viirre calculates that at these powers, the laser would need to
continuously illuminate a single spot on the retina for eight hours before any
damage occurred. 鈥淚n the VRD, the spot is travelling in two directions, and even
when it is stationary, it is not at a power that would cause damage,鈥� says
Viirre.

Where the VRD will eventually end up is anyone鈥檚 guess. But it is
likely that one of the first places the device appears will be the operating
theatre. Surgeons performing operations to remove tumours, for instance, must
avoid damaging healthy tissue that surrounds the cancer. To do this, they must
check X-rays and other scans for signs of healthy blood vessels and nerves
before they begin to operate, and wield their scalpels accordingly. But what if
the VRD could be used to project these X-ray images directly into the surgeons鈥�
eyes while they work, so that the images overlap with the surgeon鈥檚 view of the
tumour? With this kind of vital information overlaid on the real world,
previously impossible surgery could one day become a reality.

The US Air Force and Boeing have already begun their trials with the VRD, but
with so much at stake, nobody wants to comment about progress so far. But we can
be sure that right now, all those engineers, fighter pilots and partially
sighted people working with the VRD will be struggling with different parts of
the same problem. If the VRD is capable of augmenting our real world with extra
information, how will our minds handle and integrate it all? Might it
fundamentally change the way we comprehend information?