ARL Remote Identification of Suspicious Behavior and Intent of Humans

The goal of this collaboration among GTRI, VRMC, M.I.T., and the Research Laboratories of Delphi, Inc. is to identify suspicious behaviors and, where possible, to establish the subject’s intent. This will be done by using non-image based physiological measures and objective, quantifiable measures of behavior. The physiological measures will be drawn from polygraph and virtual medicine applications. The behavioral measures will be based on imaging sensors and advanced processing algorithms, which are already available in advanced vision systems developed by Georgia Tech.

A 5-step methodology will be used to develop these measures and integrate them into combined “signatures” for suspicious behaviors. First, experiments will be conducted in which human subjects act out suspicious behaviors in a social context, while the various measures are collected. In these experiments, the physiological measures will initially be collected with “contact” methods, and the behavioral measures will be interpreted by human observers. Second, stand-off or remote sensing methods will be developed for measuring the relevant physiological measures, and automatic image processing tools will be developed to replace the human interpreters. Third, the physiological measures will be combined with the image-based, behavioral measures to develop unique, multi-dimensional signatures that identify selected suspicious behaviors. Fourth, the sensors and processing methods will be tested in representative environments for several Concepts of Operation. Finally, the effectiveness of the system will be demonstrated in a working hardware/software prototype.

<< back