鈥淥OOF!鈥� Using your mouse, you heave a data file across the screen鈥攁
couple of gigabytes of data weighs a lot. Its rough surface tells you that it is
a graphics file. Having tipped this huge pile of data into a hopper that sends
it to the right program, you examine a screen image of the forest trail you鈥檒l
be hiking on your vacation. Then, using a gloved hand, you master its details by
running your fingers over its forks and bends, its sharp rises and falls. Later
you send an e-mail to your beloved, bending to the deskpad to attach a kiss.
The science of haptics (from the Greek haptesthai, 鈥渢o touch鈥�) is
making these fantasies real. A few primitive devices are extending human-machine
communication beyond vision and sound. Haptic joysticks and steering wheels for
computer games are already giving happy players some of the sensations of
piloting a spaceship, driving a racing car or firing weapons.
A more serious set of applications is supported by the FEELit computer mouse,
designed by the Immersion Corporation of San Jose, California, which will go on sale in September
(快猫短视频, 10 July 1999, p 10). You grasp it in
your hand like a conventional mouse, but electric motors provide resistance as
the cursor hits a hard barrier such as the edge of a window on the screen. The
mouse might be used by online shoppers to twang the strings of a tennis racket
or test drive a car.
Advertisement
In time, haptic interfaces may allow us to manipulate single molecules, feel
clouds and galaxies, even reach into higher dimensions to grasp the subtle
structures of mathematics.
Most of our senses are passive. In hearing and vision, for example, the sound
or light is simply received and analysed. But touch is different: we actively
explore and alter reality with our hands, so the same action that gathers
information can also change the world鈥攖o model a piece of clay or press a
button, for example. In providing direct contact between people, touch carries
emotional impact. And in providing direct contact with the world, it is the sure
sign of reality, as in 鈥減inch me鈥攁m I dreaming?鈥�
Moving finger
We know a good deal about vision, but far less about touch. Mandayam
Srinivasan, director of the MIT Touch Lab in Cambridge, Massachusetts, believes
that the complexities of touch exceed those of vision in many ways. Touch has
two different components: the pure tactile experience of moving one鈥檚 finger or
hand over a surface and feeling its qualities; and 鈥渄ynamic鈥� or 鈥渒inaesthetic鈥�
touch, sensing its physical properties of an object, such as length, weight and
orientation.
The receptors that transmit these tactile sensations are densest at the tip
of the tongue, with about 200 per square centimetre, and at the fingertips, with
half that number. They respond to pressure, vibration and temperature. Dynamic
touch also relies on information about the position of the fingers and hand,
which is gleaned from other receptors that track the extension, compression and
angular changes of muscles and tendons.
No wonder this sense is not fully understood. As haptic researcher Michael
Turvey of the University of Connecticut in Storrs notes, 鈥淚t is the tools that
are lacking. How would one quantify the time-varying deformation of our tissues,
for example, when exploring an object?鈥� The MIT Touch Lab is beginning to
address Turvey鈥檚 concerns. Their ultrasound microscope can be used to study the
mechanical and geometric deformation of the fingertip as it touches an
object.
But even without fully understanding how touch works, we can imitate the feel
of an object. The key is force feedback: when you push on a surface, it pushes
back. The exact way it responds helps you discern shape, size and texture, so a
haptic interface must mimic these forces.
Several general-purpose interfaces
are already on the market. The most successful is the PHANTOM, designed by
Thomas Massie for his undergraduate thesis project at MIT. He went on to found a
company on the device, SensAble Technologies of Cambridge, Massachusetts. In a
testing area at company headquarters, Massie introduces me to the PHANTOM. It
looks like a cross between a small multi-jointed desk lamp and a quarter-scale
dentist鈥檚 drill, ending in a pen-like stylus.
Before the computer program starts, I can move the stylus wherever I want
within a six-inch cube. But when the program is running, I can move only so far
in any direction before encountering resistance. This particular program holds
me within the walls of a small cell that contains a surprise鈥攁 shape on
its floor that I explore by feel and finally identify as the top half of a
sphere. In other demonstrations I feel weight, hardness and the shock of
collisions. The illusion is limited in what it conveys about surfaces, but it
still works amazingly well.
鈥淒igital clay鈥� is one application of this interface. Although car makers and
toy companies use computer-aided design, they still make solid prototypes out of
modelling clay because of the tactile intelligence that comes with working the
clay. But it does not integrate well with the computer-aided approach.
Massie shows me how the PHANTOM manifests electronic clay. With his left hand
working a conventional mouse, he puts a representation of a block of white clay
on the monitor, and holds it in place. Then with the PHANTOM stylus in his right
hand he sculpts the block into a convincing semblance of a bucket seat for a
sports car, using carving, gouging and smoothing tools, greatly aided by the
PHANTOM鈥檚 touch cues. When I put the gouging tool into an existing channel, feel
ing the sides of the channel helps me dig more deeply along the same line.
At $10 000 each, you won鈥檛 be seeing the device on many desktop
computers. It is a complex piece of engineering with sensors, special electric
motors and a gearless drive that smoothly and rapidly determine position in
space and generate appropriate forces, as needed for digital clay and other
applications. Even then, the device only allows the user to touch the haptic
world at a single point, not even with a whole fingertip. And the PHANTOM deals
with only three degrees of freedom, meaning it works with the x, y and z
coordinates of the hand in space, but ignores how the hand is oriented.
Delicate touch
Such refinement requires three more coordinates to describe an object鈥檚
rotation about each of its three axes: pitch, roll and yaw. Ralph Hollis of the
Robotics Institute at Carnegie-Mellon University in Pittsburgh, with his student
Peter Berkelman, has invented a joystick that is magnetically suspended,
avoiding mechanical linkages. As well as having six degrees of freedom, it is
nearly frictionless, which Hollis says makes it possible to feel delicate
forces. But at present its sphere of action is only an inch across.
In the area of human-machine interactions, Casdagay Basdogan, who works with
Srinivasan at the MIT Touch Lab, showed me how haptics can train doctors in
minimally invasive, or 鈥渒eyhole鈥� surgery. To avoid the large body openings of
conventional surgery, which leave scars and can cause infections, keyhole
surgery uses tiny incisions through which the surgeon inserts an
endoscope鈥攁 long narrow viewing device attached to a video
camera鈥攁nd special tools.
Basdogan uses two physically coupled PHANTOMs to define the front and back of
a long pair of forceps, which appear on the computer screen amidst a
three-dimensional rendering of the lungs, stomach, intestines and liver. I feel the
resilience of the tissue as I push or pull at an organ with my virtual forceps,
while watching the organ deform under the force. Basdogan notes that what I feel
and see are only approximations, as the mechanical parameters of the body鈥檚 soft
tissues are not yet well known.
This is only one of many things that must be done to improve haptic
technology. One challenge that particularly engages Srinivasan and other
researchers is to extend the haptic interface beyond the single point of the
PHANTOM, which Srinivasan likens to 鈥渢ouching reality through a stick鈥�. He wants
to make an array of tiny touch sensors and actuators that will extend over a
square centimetre.
A person can distinguish between two pencil points as little as two
millimetres apart pressed into their fingertip, so an artificial fingertip would
need that resolution to represent shape and texture perfectly. But there are
formidable engineering problems in making sensors and actuators so small using
any existing method, such as electromechanical or piezoelectric technology, and
the computational power needed to deal with this flow of data would be vast.
Haptic interfaces for a whole hand or limb may be closer to reality.
Researchers at Southern Methodist University in Dallas, Texas, have designed
what they call the Master Arm. This cyborg-like device perches on the wearer鈥檚
right shoulder, with extensions reaching down to the elbow. The Arm tracks the
motions of its user鈥檚 shoulder and elbow, and applies programmed force feedback
through a system of pneumatic pads.
Some small steps have even been taken towards whole-body haptics. Touch
Technology of Nova Scotia, Canada, has built a haptic chair. It looks like a
full-length lounge chair in a family den, but its surface is studded with 72
鈥渢actors鈥濃�攑neumatic piston rods, covered with rounded buttons, that can
extend about an inch, and can be driven under computer control in any desired
sequence and pattern. It could be programmed to imitate a real massage or to
function in time to music. According to the manufacturer, that provides a
powerful blending of sensations鈥攁 long-term goal of virtual reality.
Even at its present crude level, however, haptics can make tangible what once
could not be touched or even pictured. To investigate the world of the very
small, researchers at the University of North Carolina, Chapel Hill, have
developed the nanoManipulator. This adds touch to the technique of scanning
probe microscopy, which can image a single atom by monitoring either the
electrical current flowing between an extremely fine probe and a surface or the
force between them.
With the nanoManipulator, researchers can see and manipulate a universe a
million times smaller than their own, to study viruses and tiny semiconducting
devices. If the force feedback can be made sensitive enough, it may be possible
to push molecular keys into specific molecular locks, to custom-design drugs or
assemble silicon parts into intricate nanomachines.
With other interfaces, there is no reason we shouldn鈥檛 also be able to touch
the very large鈥攃louds, ocean currents, mantle flows, mountains, galaxy
clusters. Or the very strong鈥攚ith a suitable force scaling, new ceramics
or alloys could be squeezed and twanged to test their engineering properties. Or
the physically extreme and inaccessible鈥攕uch as ultra-hot plasma flows in
fusion machines.
Haptic technology could even make abstract ideas tangible. Many scientific
concepts occupy spaces of more than three dimensions: string theory, for
example, asserts that we live in a 10 or 11-dimensional Universe. As it is
impossible to visualise such a space, we explore these ideas through
mathematical expressions or two-dimensional sketches on paper. But probing these
unfamiliar geometries with touch may be more effective.
At the University of Tsukuba in Japan, Hiroo Iwata has designed a haptic
input that portrays five-dimensional reality. Three of the five spatial
dimensions are represented by the actual spatial location of the device, and the
remaining two by its pitch and roll. A mere flick of the user鈥檚 wrist selects
which 3D 鈥渟lice鈥� of a 5D hypercube appears on the computer monitor. Iwata tests
how comfortable users are in this 5D space by having them search for randomly
selected regions within it. When he uses force feedback to give them haptic cues
as well as visual ones, such as a twisting force that depends on position in 5D
space, their navigation through that space is noticeably improved.
And for blind people, haptics offers a new way to grasp information even in
three dimensions. A group at the University of Delaware has developed an
environment where a person can feel a mathematicalfunction. Using a PHANTOM, the
user 鈥渨alks鈥� along the surface of the figure. Like a hiker following mountainous
terrain, the user feels where the function is steep, where it is level, and
where its peaks and valleys lie. Other haptic systems could help blind people to
browse the Internet, feeling images as well as words.
Despite the fascination of extending our senses and exploring abstract
universes, many researchers see medicine as the most promising short-term
application of haptics. A surgeon could use haptic and visual information to
control surgical tools in a hospital thousands of miles away, providing instant
care for a soldier wounded on the battlefield, for example. But unless remote
surgery is the only option, patients may object to the absence of a surgeon,
diminishing as it does the element of human contact.
Other forms of contact will also be affected. Pornography is said to be the
main money-making enterprise on the Internet and to drive much of its technical
development. One can only imagine what it would mean for the cybersex business
to add digitised human touch.
It is early days, though, to think about using a computer to run your hand
over the flank of your significant other, a gorgeous stranger, or your favourite
movie star. Even the most comely digital flesh will convey little excitement if
you can only touch it with a stick. The future of haptics is bright, but the
only sensual relationship it will be sustaining any time soon is between you and
your computer.