A human-computer interface for a multi-sensor surveillance environment

Abstract

Presents a human-computer interface for a multi-sensor surveillance systems capable of adapting to the human's changing performance with time. The interface's algorithm works as follows. As the human performs tasks, his performance is measured and this knowledge is integrated together with the estimated task arrival and service rates. An optimal task allocation, based on the current human and system performances, is chosen which then determines which tasks are allocated to the human and to the computer respectively. The algorithm was tested within a human controlled surveillance system. Results showed that the use of the interface enhanced the human's performance during periods of high mental workload