Teaming Human and Artificial Intelligence for High-Productivity Security Systems,
Abstract
Multi-sensor security systems which place primary emphasis on high probability of detection (PD) of remote equipment only address one third of the overall security problem, which is to pose a high probability of interdiction. The other two critical steps are communicating information from the machine to the security console operator (reporting), and dispatch of personnel (response). Lapses in either of these latter two functions can easily defeat the goal of near-certain interdiction. For example, a system which depends on a human operator's sustained attention to a wall of CCTV monitors for eight hours does not adequately address the reporting problem. Systems where alarms are pouring in too fast to all be responded to are underplaying the response issue. Throwing more security personnel at the problem does not appear to be an acceptable solution. Recent artificial intelligence (AI) progress in computing situation assessments directly from remote sensor inputs suggests a method for weeding out nuisance-type false alarms in multi-sensor security systems by building in more discriminability. This approach also fosters reporting flexibility at the man-machine interface, making it adaptable to event traffic density. A total systems approach, employing remote sensors, intelligent processing, a well designed man-machine interface, and trained security personnel offers a well-balanced teaming of technology and manpower to produce high-deterrence, cost-effective security in coming years. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 16, 1982
- Accession Number
- ADP001597
Entities
People
- Pierre Bierre