Human-Animal Discrimination Capability for Unattended Ground Sensors

Handling false alarms produced by animals is a special problem for unattended ground sensors (UGS) because of the number of ways that animal signatures can mimic those of humans. Specifically, animal footsteps are impulsive, much like humans; the contact of a foot or hoof on the ground produces a sudden noise that rapidly fades away. Seismic, acoustic, and passive infrared (PIR) sensors are often plagued by false alarms and this often results in the distrust of such systems. Toyon Research Corporation proposes to develop a signal processing framework to eliminate false alarms produced by animals and allow for the detection of humans. Toyon's approach consists of target detection and feature extraction modules that are specific to each sensor modality, namely seismic, acoustic, and PIR. Next, automatic target recognition (ATR) algorithms will be applied on select features to conduct preliminary discrimination for each modality. Finally, a fusion algorithm will combine the ATR discriminations from each of the three modalities and output the consensus, high-confidence target declarations. Toyon's efficient processing structure allows for a low-power, real-time solution that will be agnostic to UGS hardware systems. Toyon's approach can easily be integrated with existing DHS and DoD platforms.