Abstract
Many contemporary human-machine systems are becoming increasingly automated. In some instances this automation is complete (robot-operated assembly lines). The Japanese automobile industry is a prime example of this type of implementation. In 1965 approximately 68 working hours were required to produce one subcompact automobile. Automation of press forming of sheet metals, casting, forging, welding, and engine assembly had reduced that number to approximately 25 by 1977. The military has also followed the trend towards automation with the development of the Patriot missle system, designed to identify targets, make engagement decisions, assign weapons, and lock on to and engage the target, all without human intervention.
Other systems have been designed at the semi-automation level, allowing major portions of the overall task to be performed by machine/computer subsystems. This is evidenced in petrochemical plants where some control loops are closed and others are left open, requiring some intevention on a continuous basis by the operators. This same situation exists in airborne flight control and navigation systems. Although many of the subtasks are fully automated, the overall task requires some human intervention for successful completion.
Reliability of these complex systems is less than perfect, however. In 1977 a test of the Worldwide Military Command and Control System (WMCCS) indicated a 62% failure rate. In 1980 the avionics systems on the F111-D managed to endure for an average of only 3 hours between failures, requiring an average of 33.6 maintenance man-hours per sortie. Perhaps one of the most recent, and memorable, examples was the episode during which the Space Shuttle onboard computers had such a serious argument that they refused to speak to one another…
In some systems this eventual failure, ranging from slight performance decrement to catastrophic malfunction, can be dealt with by shutting down the system and affecting repairs. This luxury is usually not available in air transport where airborne failures require manual intervention for successful task completion. If the operator of a system has been removed from continuous or semi-continuous control for too long, serious degradation in skills is likely. Evidence of this problem has been observed in aviators; first officers qualified in a highly semi-automated aircraft experience difficulty when promoted to captain in an aircraft lacking these semi-automated systems. Similar problems have been observed in the petroleum industry when computers providing advisory information fail, forcing the operators to make production and process adjustments unaided. The operators, in both instances, have become dependent upon the automated subsystems through disuse of skills. Thus it would seem reasonable to assume that the quality of human intervention would be positively correlated with frequency of intervention, suggesting that the more reliable systems should suffer significantly during operator interventions.
The three major operational skill categories involved in aircraft pilotage, continuous manual control, procedural control, and communications, should be examined in relation to three major areas of concern. First, it is necessary to determine how rapidly these skills deteriorate when unused and what factors (previous learning, skill category) influence that deterioration. Second, one must discover what measures may be effective in the prevention or reduction of such deterioration. This includes assessment of simulator utility, both part- and whole-task. Finally, means must be devised to assess the quality of skill retention in order to assure that the implementation of skill-retention strategies continues to be effective. The goal of the current program in this area is to establish task, training, and system design criteria that will allow the system operator to maintain a high level of proficiency without imposition of elevated workloads or meaningless “busy-work” tasks.
Get full access to this article
View all access options for this article.