In a presolicitation posted earlier this month to the Federal Business Opportunities website, DARPA makes an unusual request:
DARPA seeks the comprehensive acquisition and analysis of atmospheric trace gases present in selected urban areas, along with concentration modeling to increase the spatial and temporal resolution.
Ok, so DARPA is looking for new ways to acquire and analyse data about gases in cities. That sounds relatively tame, and even a bit mundane. Until you look at why they want this data:
It is expected that [analysis of this data] will provide anomaly detection, source attribution, efficient sampling strategies, and the generation of actionable intelligence.
This presolicitation constitutes at least one prong of a system that could--or rather will--be used by law enforcement, and in particular, for counterterrorism. DARPA's hope, according to this posting, is that with the right equipment and data processing software, it will one day be possible to map an entire city based on its particular odors in real time. Or something close to it.
This call for proposals is just one of many that will be made before large-scale scent monitoring comes of age, and focuses mostly on refining one initiative, the Chemical Cartography: Urban Data Acquisition effort. The goal of this effort is to figure out how to create "three-dimensional high resolution spatial and temporal chemical composition ground-truth maps" of urban environments. In other words: smell maps. This is neither easy nor cheap.
DARPA's document assumes chemical cartography data samples coming in through various arms of a Chemical Reconnaissance program at a rate of 300,000 a day, eventually. The problem is, this data would be meaningless without proper analysis. The crux of the problem? Urban environments don't just smell weird; they smell weird in changing ways.
Behavior of the diverse chemical species within an urban environment adds a layer of complexity to the chemical map. Chemical properties will lead to different transport, dispersion, and interaction patterns within the environment. Thousands of chemicals will be present within an urban environment. Accurately characterizing their movement and interaction within this environment must be addressed to create a realistic urban chemical map.
Anyone who's been to New York or San Francisco in the summer can attest to this: there's a lot going on in our nostrils at any given time, and it's not always clear where each smell is coming from.
But even creating a map that designates the presence all these different fumes isn't enough. DARPA hopes to use this data for "anomaly" detection, which means determining whether the presence of a chemical in the air is explainable or appropriate in a specific context, at a specific time. (The example DARPA invokes is of diethyl ether, which would be considered innocuous if detected near an engine shop, but if sensed near a suspected drug den might point to the production of freebase cocaine.)
For this, DARPA needs people to come up with smart, predicitive models, which take into account all the weird emissions caused by a variety of different urban fixtures--gas stations, restaurants and cleaners, to name a few--and can extrapolate them across an entire city.
Once these models are established, a man or vehicle with simple chemical sampling canister would be able to sniff out an area near to a suspected bomb manufacturing facility, or drug production facility, or even a legal facility that handles dangerous chemicals, and quickly know if the area is any "smellier" than its environment--local businesses, homes and climate--says it should be. (Hat tip to Wired)
Image taken from the Jason Logan's hilarious "Scents and the City" map, published on the NTY website