Military Embedded Systems

DARPA surveillance system uses EEG for image filtering

News

September 20, 2012

Brandon Lewis

Technology Editor

Embedded Computing Design

DARPA surveillance system uses EEG for image filtering
Photo credit: Advanced Brain Monitoring

ARLINGTON – The Defense Advanced Research Projects Agency (DARPA) has released field test results of the Cognitive Technology Threat Warning System (CT2WS), a threat detection system utilizing a camera that interacts with operator brain waves to scan wide areas at extended distances. Test results revealed a 91 percent success rate in threat detection using the CT2WS, compared to the 53 percent successful detection rate of more traditional camera systems.

The CT2WS is composed of a tripod-mounted, high-powered, 120-megapixel video camera with a 120º field of vision that works in conjunction with a real-time brain signal monitoring electroencephalogram (EEG) cap to that transmit signals to a computer system with image processing software. The system then presents an operator with up to 10 images per second, and by leveraging his or her P-300 brainwave (a wave associated with the audio and visual processing mechanisms of the brain) and cognitive visual processing algorithms, it then highlights images of interest for further review, which are indicated by spikes in brain activity.

Using just its own visual processing algorithm, the system produced 810 false positives out of 2,304 “target events, whereas when a human operator’s brain was leveraged in tandem false alarms fell to five per hour. When a portable radar was included in the system as well, 100 percent detection was achieved.

Developed by a team including the Quantum Applied Research, Advanced Brain Monitoring, the University of California, San Diego and HRL Laboratories, work on the CT2WS began with a Pentagon initiative in 2007.

 

Featured Companies

U.S. Defense Advanced Research Projects Agency (DARPA)

675 North Randolph Street
Arlington, VA 22203-2114
Categories
A.I. - Big Data
Topic Tags