The performance of a human-machine system can be influenced by the type of interface that is used to control the dynamic allocation of functions between human and machine. Of particular interest is the performance of the human around changes in function allocation. Two experiments are described which examine the allocation of function between manual and automated control. Using a primary flight-type tracking task, which could be automated or controlled manually, with two possible secondary events - identification of a target shape, and response to an abnormal gauge - the research studied the effects on human performance of control mode switches. The first experiment examined the influence of feedback type - either visual, direct voice output (DVO), visual and DVO, or none - when the control mode switch was automated, while the second experiment examined the influence of the response type - manual, direct voice input (DVI) or none - when the operator was prompted to make a control mode change. A considerable increase in tracking error was found, lasting for several seconds around the moment when allocation of function changed, regardless of the direction of the change in automation. Although the effects were transient, the magnitude and duration of the phenomenon increased with increased task loading, and it was not the case that DVO/DVI technology was inherently able to overcome these effects. While the magnitude and duration of the transient tracking errors was system dependent, these findings may suggest hazards of operational significance, both for present and future automated systems. (A)
Abstract