DARPA Has $3.2M to Sniff You Out 223
quackking writes "The Army wants to sniff you out. This fedbizopps.gov link to a DARPA pre-RFRQ tells more. 'The Defense Advanced Research Projects Agency (DARPA) Advanced Technology Office (ATO), as part of the Odortype Detection Program, invites proposals to (1) determine whether genetically-determined odortypes can be used to identify specific individuals, and if so (2) to develop the science and enabling technology for detecting and identifying specific individuals by such odortypes. Total program funding for this effort will not exceed $3.2 million in FY 2003. Multiple awards are anticipated. Proposals are due by January 29, 2003.'"
Not possible (Score:2, Informative)
Body odours are, as the proposal points out, the result of carboxylic acids.
Although the term carboxylic acid covers very many molecules - basically anything with a HO-C=O somewhere on it, the molecule has to be volatile to have a smell. The problem is that not many acids are volatile - the very composition of the molecule means it makes hydrogen bonds with others easily, and even light acids are involatile liquids or solids.
This means there is a small pool of molecules to pick from so the chances of an individual having a unique blend is very small.
Re:Sounds like a waste of 3.2m (Score:3, Informative)
The theory on how animals tell people apart has to do with combinations of long-chain carboxylic acids. They have long fatty chains, and the long carbon chains keep the volatility low, but dogs can supposedly detect them. Their noses are quite sensitive for these compounds; much more so than human noses. On the other hand, humans noses are more sensitive for some other compounds... the mercaptans, for instance.
I just wonder if they can make equipment that will detect such low, low levels of these compounds, and whether there is enough variation (with the limited number of long chain fatty acids) to produce a unique signature among billions of people.
Unless I'm way off-base and they are going in a totally different direction, I don't see how this is possible.
Aromatic Compounds (Score:4, Informative)
Re:Intresting stuff (Score:4, Informative)
Dog and cat info (Score:2, Informative)
Here's a link that gives a somewhat simplified explanation of dog and cat taste/smell... it even spares you some of the arcane aspects of organic chemistry
Animal senses [speedyvet.com]
Re:Why (Score:4, Informative)
Tho if this becomes practical, I foresee a thriving market in whole-body deodorant washes. It won't fool a really good dog (probably because they're sensitive to a whole spectrum of scents) but I'd bet it would fool a sensor-and-database arrangement, which perforce would be more limited in its sensory range.
BTW, German Shepherd Dogs have poor noses compared to Labrador Retrievers, and Labs train up easier for this sort of work.
Re:Intresting stuff (Score:3, Informative)
Scent masking (such as with ordinary perfume) doesn't work very well against dogs, because an experienced dog will pick up on the secondary or combined scent and follow that instead. Plus they aren't looking for just one particular scent molecule, but whatever combination the target happens to exude.
similar research (Score:3, Informative)
It's probably just a matter of adapting the existing technology. It will be interesting to see if they can distinguish one scent out of say a bunch of people in an airport looking for a known criminal.
Re:Sounds like a waste of 3.2m (Score:3, Informative)
Well, to be fair it was a 17-year-old medium-format, professsionally archived head shot photo. That is why there was enough information preserved in the negative to make an iris match. That won't work with just any old 35mm pic.
smells like teen spirit (Score:2, Informative)
great sig. actually, this sort of thing may be doable. the DARPA web solicitation refers to sampling the environment for a chemical signature derived from a person's Major Histocompatibility Complex (MHC) proteins. these have a genetic basis and when sampling blood can be used to discriminate individuals without examining the DNA which encodes them. obviously, the technology DARPA wants developed will be to sample small amounts of protein from a defined environment to look for an immunological signature of an individual.
from a distance and without making direct contact with an individual, this type of technology would obviously be more useful to remote monitoring devices than fingerprints, DNA, etc. who knows, possibly in a 2nd or 3rd gen incarnations something like this could be used in a predator type aircraft scanning an area. flying bloodhounds in a sense.