Standoff technologies for the detection of Explosively Formed Penetrators (EFPs)

The rise of the explosively formed projectile (EFP) as a tool of insurgents is indicative of the enemy?s resolve and willingness to exploit technologically advanced munitions against US forces. Detection and classification of roadside EFPs is difficult due to the complicated nature of EFP responses and the cluttered nature typical of roadside environments. This proposal, in response to Army SBIR A14-039, aims provide route-clearance capability for roadside EFP threats by 1) leveraging existing prototype sensors already under development (synthetic aperture acoustics, SAA; and forward-looking ground-penetrating-radar, FLGPR), and 2) making use of state-of-the-art advances in machine learning, feature extraction, and physics-based modeling for explosive threat detection. In particular, this work will explore responses in SAA and FLGPR data from both EFP and non-EFP objects, and identify the distinguishing aspects of each. These features will be automatically extracted from SAA and FLGPR data using approaches originally proposed in the computer vision literature. Classification of threats from non-threat responses will be accomplished using advanced pattern recognition approaches, and the robustness of the approaches will be explored using physical models and robust cross-validation. The proposed effort will result in the first generation of prototype algorithms suitable for real-time roadside EFP detection.