Academic journal article Human Factors

A Quantitative Measure for Degree of Automation and Its Relation to System Performance and Mental Load

Academic journal article Human Factors

A Quantitative Measure for Degree of Automation and Its Relation to System Performance and Mental Load

Article excerpt

INTRODUCTION

Automation has been employed in industrial processes to achieve consistent product quality, minimize operator workload, reduce human error during the execution of repetitive tasks, improve process availability and production efficiency, and enhance operational safety. Highly automated systems have become available as a result of developments in computer and information technology. These technological achievements have changed the role of human operators and the nature of the operator's task during the operation of complex systems. The role of operators has shifted from manual to supervisory control - that is, from performing "doing" tasks, such as activating manual switches and following an operation procedure, to performing intellectual or cognitive tasks, such as diagnosis, planning, and problem solving (Hollnagel, 1995).

However, a high level of automation may have adverse effects. Bainbridge (1983) discussed the possible consequences of increasing the level of automation. In the context of human-machine systems, automation is applied only to control tasks that are fully understood and that are completely described. The tendency has been to automate the easiest tasks and to leave the rest to the operator. From one perspective this dignifies the human contribution; from another it may lead to a hodgepodge of partial automation, making the remaining control tasks to be performed by the operator less coherent, less meaningful, and more complex than they need be (Sheridan, 1992, p. 358).

The human operator may also have difficulty performing those tasks during situations that are not completely understood or foreseen by system designers (Rouse, Cody, & Frey, 1992). This can result in an overall degradation of system performance. Furthermore, higher levels of automation imply higher system complexity, which affects both operator performance and system performance (Stassen, Andriessen, & Wieringa, 1993). Automation may also increase the complexity of some human tasks.

Increased automation, therefore, does not necessarily result in increased benefits. In fact, too much automation results in poor operator performance caused by too-low workload, loss of skill, and loss of awareness of the system status. This raises a series of questions: How much automation should be used? What affects human use of automation? How does an increase in level of automation affect system performance and operational safety? Many such questions assume the existence of some quantitative measures for the degree of automation. To our knowledge, no such measure has been developed.

Level of automation, as a qualitative concept, has been used in the literature to indicate how much automation is included in the design (Billings, 1991; Endsley & Kids, 1995; Kantowitz & Sorkin, 1987; Sheridan, 1992). Sheridan's (1992) scale classified the level of automation according to the distribution of responsibility (task allocation) between human and computer (Johannsen, Levis, & Stassen, 1994; Levis, Moray, & Hu, 1994). He determined the level of automation by comparing the types of tasks performed by a human (such as decision making, execution of control tasks, and monitoring) with the types of tasks performed by automation (computer). However, the number of these tasks performed by the human versus the automation was not taken into account. Similar to Sheridan's scale, a scale of the level of system control and management automation was discussed by Billings (1991) and Endestad (1991).

To supervise a system, the operator or automation has to perform a number of subtasks associated with the main supervisory task, such as set-point control, monitoring, diagnosis, and decision making. However, the qualitative scales just discussed do not take into account the number of tasks performed by the human and those performed by automation. A quantitative measure for the degree of automation should take into account not only the different levels of automation that exist but also the different natures and number of tasks. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.