Great logo -- kind of sets Cyrano Sciences up for being bought out by
Time Warner's eye-and-ear :-)
I trust you had a pleasant vacation since our encounter at the Avis
desk in Kona. In case you hadn't seen it, this was the Economist
article I was referring to.
Rohit Khare '95
Artificial noses. Nowt to sniff at
Wine-tasting will never be the same again
DUPLICATING human senses is a tricky engineering problem. Anyone
wishing to do it must imitate both the organs that capture sensory
information and the brain cells that interpret it. Workers on
artificial vision and hearing have the advantage that the sensory
parts of their equations-cameras and microphones-have been around for
ages. Makers of artificial noses, however, must start from scratch.
The possible applications of an artificial nose are legion. Medical
diagnosis (which relies more on a doctor's sense of smell than most
patients realise) could be speeded up. Landmines could be detected
from the odour of the explosives they contain. Vineyards and
breweries could have the quality of their products monitored on the
fly. And a sniffer located over the scales at a supermarket checkout
could tell the cash register what sort of fruit or vegetable was in
the pan, and therefore how much to charge. According to some
analysts, the world market for these types of devices would be worth
about $4 billion a year.
Such artificial noses as now exist usually rely on electrically
conducting polymers (long, chain-like molecules) to duplicate the
receptor proteins inside a real nose. There are about 1,000 different
receptor proteins, each of which responds to a handful of odoriferous
molecules. The brain is able to integrate the signals these proteins
give out in response to particular mixtures of molecules and
recognise the substance as, say, coffee.
By making a series of detectors out of various polymers which react
differently to particular mixtures of molecules, it is possible to
mimic this arrangement-though with many fewer receptors. Its reaction
with the odour molecules changes a detector's electrical resistance,
and this signal is fed into a neural net (the closest that computer
technology has got to imitating the pattern-recognising abilities of
networks of nerve cells) to work out what is being smelt. Only a few
polymers, however, conduct electricity, and so the range of chemicals
that can be detected is limited to those that will react with those
At the California Institute of Technology (Caltech), Nathan Lewis is
building on the idea of conductive polymers by adding carbon black
(in essence, soot) to the polymer. This provides the necessary
variety of response because the task of conducting the electricity is
taken over by the carbon black. The resistance of a detector changes
as the polymer swells in response to its reaction with the odour
molecules and thus alters the connections between the particles of
carbon black. That means that a whole range of polymers which do not
conduct electricity (such as most common plastics) can be brought
into play. With a big enough range, almost any airborne molecule will
react with at least one of them.
Dr Lewis's current prototype has 17 sensors. The result, when fed
into a computer, is a 17-dimensional picture of whatever scent is
blowing over these sensors. With the aid of some mathematical
finagling this picture can be reduced to a mere three dimensions, for
display to a human operator. Once the computer has learnt what's
what, it can recognise and compare smells, or can give its human
operator the chance to do so. If two scents are closely related-a
designer perfume and a cheap impostor, for example-their profiles
will be similar, but not identical, and the pictures will be visibly
different. As Dr Lewis puts it, pure-polymer detectors see only in
shades of grey; the carbon-black composites, with their vast range of
polymers, provide the olfactory equivalent of colour vision.
The other new technology in development uses colour in a rather more
literal sense. The so-called "optical nose" is being built by John
Kauer and David Walt, at Tufts University, near Boston. In one
respect-its reliance on the absorptive qualities of plastics-it is
similar to the design used by Dr Lewis. But instead of producing an
electrical signal when it is tickled, the optical nose sneezes light.
The nose's sensors consist of microscopic plastic beads coated in
fluorescent dye and perched on the ends of optical fibres. Each bead
is doped with a mixture of chemicals that become electrically charged
in response to odour molecules-and that charge alters the colour of
the dye. By doping beads with different mixtures, a range of
responses to a given combination of odour molecules can be obtained.
The changing colours are transmitted through the optical fibres to a
camera, and thence fed into a neural net for analysis.
Naturally, both research groups have their commercial arms. The aptly
named Cyrano Sciences, based near Caltech in Pasadena, plans to put a
hand-held device based on Dr Lewis's technology on the market in July
1999. Illumina, located further south in San Diego, hopes to have a
commercial version of the optical nose ready about a year later.
Perhaps a blind test of the nose of the world's best wines could then
decide which technology is the true master.