Safety Dynamics' acoustic recognition technology requires the "learning" of specific sound signatures to operate effectively. Sound signatures are currently being researched, recorded, and tested for possible future use.
DSNN has great potential for effective recognition of sounds that would allow for protection in the areas of pipelines, forests and facilities.
At the perimeter of a facility, whether in an urban environment or a non-urban one, SENTRI smart sensors can be trained to recognize the sounds of a threat to a military base, a government building, or a public venue and can always be trained to recognize new sounds.
Currently, sound signatures are being recorded and tested for engines and chain link fence intrusion. These are being designed and trained to work for perimeter defense and protection, and are visualized to work in concert with other intelligent sensors and cameras.
These signatures are in development for future use and are not currently available.
The SENTRI Solution: A New Age In Law Enforcement
Safety Dynamics specializes in the use of smart sensors for threat recognition and localization. Safety Dynamics is currently selling and supporting a system for law enforcement called SENTRI (Sensor Enabled Neural Threat Recognition and Identification). The system is a breakthrough technology that recognizes gunshots and explosions and sends range and bearing details to cameras which can then locate the source of the event.
The patented Dynamic Synapse Neural Network (DSNN) technology developed by the Laboratory for Neural Dynamics at the University of Southern California is at the core of the acoustic recognition capability and is based on neurobiological principles of brain signal processing, and allows, like the human brain, accurate temporal pattern recognition of acoustic signals even in the presence of high noise.
The SENTRI DSP Core
Identification of acoustic signals in noisy environments remains one of the most difficult of signal processing problems, and is a major obstacle to the high degrees of accuracy and speed needed to identify suspicious sounds in high-security, high-safety environments.
With these fundamental concepts in mind, Safety Dynamics has developed a fundamentally different paradigm for biologically-based pattern recognition, which incorporate nonlinear dynamics of neurons in the hippocampus, the brain region responsible for forming pattern recognition memories.
The small footprint of a DSNN-based classification system makes DSNNs highly desirable for use in field-deployed devices and Safety Dynamics implemented its first commercialized DSNN in the form of a gunshot recognizer. Further development in under way to train the DSNN for a library of sounds associated with threats and intrusion, giving Safety Dynamics the ability to expand its acoustic recognition to sounds beyond gunshot, while retaining and improving the perfomance of recognition in high noise envrionments.
Safety Dynamics' SENTRI at the Cisco Network Demonstration Facility in San Jose, CA.
SENTRI employs the trained DSNN technology gunshots to recognize small arms and assault rifle fire and not to respond to other loud noises, like a city bus backfire.
The technology is designed to identify gunshot events in noisy environments, determine the location/origin of the gunshot, direct a video camera to collect data on the gunshot source, and wirelessly send a signal to a central command site, where officers can immediately begin monitoring the video stream.
SENTRI: Single Unit or Network Grid
SENTRI acoustic recognition is currently being installed for the purposes of urban crime prevention. Whether standing alone in a choke-point or working with mulitple units to cover a large area, SENTRI is part of a network of surveillance cameras which listen for gunshots and provide police with the ability to use audio and video for the identification of crimes in progress.
SENTRI with Video and Directional Analyzer
Perimeter Security
Safety Dynamics is currently developing a perimeter security prototype under funding from the Navy. This research involves training the DSNN to recognize sounds related to security breaching and integrating other sensors to create a complete perimeter defense solution.
This effort has included training the DSNN to listen for vehicle engine sounds, footsteps, climbing and rattling of chain-link fencing, and will include an interface to ObjectVideo's Intelligent Video product.
A possible design for perimeter security and protection of fixed assets.