Human performance during automation: The interaction between automation, system information, and information display in a simulated flying task.

摘要

A single experiment is reported that used a multi-attribute task battery to simulate a flying task to determine the effects that automation had on human performance. The task battery consisted of three tasks: system monitoring, resource management, and a dual (primary and secondary) tracking task. The research was conducted to determine the role of automation management and an automation hierarchy on performance with the task battery. The automation hierarchy used included fully manual, management by delegation, management by consent, management by exception, and fully automatic automation engagement. This hierarchy is consistent with components of Human Centered Automation. These components were assessed to determine whether they caused reduction in operator involvement, awareness, and performance.;Two additional hierarchies were investigated: Information Management and Information Display. Information Management was functionally defined by varying the nature of the information presented in support of the resource management task. Raw data, status, and recommendation were evaluated to determine whether the type of information presented for a task affects the involvement of the operator. Information Display was functionally defined by varying the nature of the interface to provide two formats: dedicated and integrated.;The results of the study are evaluated from the perspective of Human Centered Automation. When appropriate, findings are evaluated using Multiple Resource Theory. The usefulness of each theory in accounting for the experimental results is discussed. The results indicate that Automation Management (AM) does not follow the hierarchical nature described by Human-Centered Automation. Additionally, rather than cause peripheralization, use of automation resulted in enhanced performance. This performance benefit is more consistent with a reduction in task load than a shift in processing resources. In fact, the research suggests that a multiple resource theory does not adequately explain the cause of shifts in performance due to automation. The use of raw data during automation transition resulted in deficits in performance compared to use of status or recommendation information. Conversely, use of a recommendation to prompt operator performance may be limited by the discrete nature of the annunciation. Finally, use of schematic representations in display formats had no effect on performance.