HEMA - a Robust, Computational Architecture for Modeling Human Error

The growing use of sensors and other data-generating devices increases the chance of operator error. This is especially true of cutting-edge weapons systems, where the human operator must receive, process and act on an increasingly-complex array ofinformation from numerous sources. Human error in these systems can lead to serious and even deadly consequences. A tool that can help predict the occurrence of these errors early in the system design process would be extremely valuable to both systemdesigners and the human operators of these systems. Such a predictive tool would assist the system designers to minimize both the incidence and severity of human errors. D.N. American proposes the development of this predictive, human error modelingtool. The Phase I team brings substantial expertise in cognitive psychology, cognitive architectures, simulations, genetic algorithms and parallel programming to this approach. This combination of expertise will ensure the success of this innovativehuman error modeling tool. The initial target end users of this software tool will be current suppliers of information systems, software/hardware systems, etc. to the Department of Defense. However, the trend of increasing the amount, speed and formatof data delivered to human operators is evident in numerous civilian applications. Examples include air traffic control operators, manufacturing control centers, automobiles, call centers, etc. In each of these examples, the proposed human error modelingsystem will help reduce the incidence and severity of human errors. Consequently, the proposed technology will prove valuable to designers of system that involve human/machine interface in these civilian applications as well.