Human Performance Effects of Adaptive Automation of Variuos Air Traffic Control Information Processing Functions
No Thumbnail Available
Files
Date
2004-02-15
Authors
Journal Title
Series/Report No.
Journal ISSN
Volume Title
Publisher
Abstract
Advanced forms of automation are being considered for application to Air Traffic Control (ATC) in order to reduce controller workload and make higher traffic volumes more manageable. At this point in time there is only limited knowledge of the implications of, for example, Adaptive Automation (AA) on controller performance, workload, and Situation Awareness (SA). The purpose of this research was to (1) define a measure of SA in an ATC simulation that is a sensitive and reliable indicator of automation state changes as part of AA; (2) empirically assess the SA measure for use in investigating AA of various ATC information processing functions; and (3) determine the relationship(s) between AA in the ATC simulation and operator SA. The situation awareness global assessment technique (SAGAT) was considered to be an appropriate candidate measure of SA during AA of an ATC simulation.
An experiment designed to empirically assess the sensitivity of a SAGAT-based approach included an ATC simulation, Multitask©, which is capable of simulating five forms of ATC information processing automation: manual control (no automation), information acquisition, information analysis, decision making, and action implementation. A secondary, signal detection-based gauge-monitoring task was used to measure Multitask© workload and trigger dynamic function allocations (DFA) between manual and automated control. Eight subjects were recruited for the study and each performed under the five modes of automation twice (one repetition). Performance measures were collected for both the Multitask© simulation (aircraft cleared, conflicts, and collisions) and the gauge-monitoring task (workload).
The SAGAT involved freezing the simulation three times per trial and querying the subjects on their perception, comprehension, and projection (Level 1, 2, and 3 SA) of simulation states. Subject responses were scored based on ?ground truth? observations recorded during stops. Data was analyzed, first, on a per stop-basis in order to identify any effect of the general type of control (manual or automated), and then on a per trial-basis to identify any mode of automation effects.
Results revealed that AA of the action implementation aspect of ATC information processing produced superior performance during the automated control periods as part of the simulation. In addition, automation of the information analysis aspect of ATC information processing appeared to have a positive effect on performance during manual control periods as part of adaptive conditions. Subjects were able to benefit from the automation even during periods of manual control of the simulation. Counter to expectations, no significant findings were revealed for secondary task performance (workload). Situation awareness results revealed that subjects were better able to project the future simulation status (Level 3 SA) during manual control periods of the simulation than when operating under automation. The SAGAT data analysis revealed no significant effect of the specific mode of automation on SA. In general, these results suggest that SAGAT was not a sensitive measure of SA in the simulated ATC task. This was primarily attributed to a lack of consideration of the relevance of particular aircraft to a controller, at any given point in time in the simulation, as part of measurement technique. As a result, the final objective of determining relationship(s) between AA in ATC and controller SA was not achieved.
The performance results of this study were in agreement with the results of prior research demonstrating ATC information processing to benefit most from lower-order automation, like action (clearance) automation. The study also further demonstrated the use of a secondary-task measure of workload as a basis for facilitating AA in a complex task.. With respect to SA, the study exposed a need for future research to assess the sensitivity and reliability of SAGAT in an ATC environment and the potential for development of a more sensitive SAGAT-based measure for this domain that considers issues of aircraft relevance and concurrent controller goal sets.
Description
Keywords
adaptive automation, SAGAT, situation awareness, air traffic control, human information processing, dynamic function allocation, automation
Citation
Degree
MS
Discipline
Industrial Engineering
