@CONFERENCE\{IMM2015-06908,
    author       = "J. K. Munk and O. T. Buus and J. Larsen and E. Dossi and S. Tatlow and L. L{\"{a}}ssig and L. Sandstr{\"{o}}m and M. H. Jakobsen",
    title        = "CRIM-TRACK: Sensor System for Detection of Criminal Chemical Substances",
    year         = "2015",
    month        = "sep",
    keywords     = "Trace vapor sensing, drugs, explosives, precursors, colorimetry, optics, disposable chip, user friendly",
    booktitle    = "Proceedings of {SPIE} Security + Defense,  September 21-24, Toulouse, France",
    volume       = "9652",
    series       = "",
    editor       = "",
    publisher    = "SPIE",
    organization = "",
    address      = "",
    note         = "Session: Detection and Identification of Things Dangerous, Hidden and Suspicious {II}",
    url          = "http://www.spie.org/sd",
    abstract     = "Detection of  illegal  compounds  requires  a  reliable,  selective  and sensitive  detection device.  The  successful  device features automated target  acquisition,  identification and signal  processing.  It  is portable,  fast,  user friendly, sensitive,
specific, and cost efficient. LEAs are in need of such technology. {CRIM-TRACK} is developing a sensing device based on these requirements. We engage highly skilled specialists from research institutions, industry, SMEs and LEAs and rely on a team of end users to benefit  maximally from our prototypes.  Currently we can detect  minute quantities of drugs,  explosives  and precursors  thereof  in laboratory  settings.  Using colorimetric  technology we  have  developed prototypes that employ disposable sensing chips. Ease of operation and intuitive sensor response are highly prioritized features that we implement as we gather data to feed into machine learning. With machine learning our ability to detect threat compounds amidst harmless substances improves. Different end users prefer their equipment optimized for their specific field. In an explosives-detecting scenario, the end user may prefer false positives over false negatives, while the opposite may be true in a drug-detecting scenario. Such decisions will be programmed to match user preference. Sensor output can be as detailed as the sensor allows. The user can be informed of the statistics behind the detection, identities of all detected substances, and quantities thereof. The response can also be simplified to “yes” vs. “no”. The technology under development in {CRIM-TRACK} will provide custom officers, police and other authorities with an effective tool to control trafficking of illegal drugs and drug precursors."
}