Performance and workload effects for full versus partial automation in a high-fidelity multi-task system.

摘要

This thesis evaluated theoretical predictions concerning performance and workload effects of the implementation of adaptive allocation. Five experiments are reported in which adaptive allocation was implemented in a multiple task aviation simulation with component tracking, monitoring, and target identification sub-tasks. Experiments 1 and 2 empirically determined input values for the tracking task which produced controlled levels of tracking difficulty. Experiment 3 exposed pilots and non pilots to single, dual, and multiple task combinations under independent and linked sub-tasks configurations. Results indicated that performance on all sub-tasks was sensitive to the number of concurrent tasks and further indicated that the non-linked system configuration contributed to reduced pilot efficiency. Experiment 4 implemented adaptive allocation for the tracking sub-task based on a model which identified an increase in tracking error during the initial presentation of a target. During initial target presentation, tracking control was either fully or partially allocated to the system for a brief period, after which full control was returned. Results indicated performance benefits on all tasks for both full and partial adaptive allocation strategies and confirmed that an independent task configuration may underestimate pilot efficiency. Experiment 5 extended the implementation of adaptive allocation to include adaptive display layout. It compared functional grouping of sub-tasks displays based on principles of perceptual and processing proximity. Results provided support for the implementation of adaptive display design in general but failed to support the specific layouts derived from the proximity conception.