A study was conducted to determine whether air traffic control (ATC) communication events (number and duration of controller/ pilot communications) would predict subjective estimates of controller workload as well as taskload measures based on aircraft and controller activities. Analyses were conducted that compared different regression models' predictions of subjective workload estimates made by 16 subject matter experts for 8 samples of air traffic activity. The predictors in the regression models were different combinations of five taskload principal components computed from routinely recorded ATC data and two measures of pilot/controller voice communications. A series of model comparisons was conducted to determine whether a "reduced" regression model containing fewer variables would predict the workload ratings as well as the full model containing all predictors. Several reduced models predicted ATWIT ratings as well as the full model, but a reduced model containing only the communications variables was not as effective. The results suggest that certain voice communications measures add nothing to the prediction of subjective workload, over and above that of taskload.
Two computer programs, the National Airspace System (NAS) Data Management System (NDMS) and the Performance and Objective Workload Evaluation Research (POWER) program, have been developed to provide a platform for quantifying en route air traffic controller activity and taskload. The NDMS program extracts data produced by en route mainframe computers and encodes the information into database files that provide efficient storage and access. The POWER program calculates specific measures using aircraft positions and controller data entries. The development and use of such measures is important for establishing baseline activity measures and for evaluating modifications to ATC systems. NAS System Analysis Recording (SAR) data were collected from the Jacksonville en route air traffic control center between 8:30-10:30 a.m. and between 12:00-2:00 p.m. (local time) for each of four consecutive days. POWER measures were computed in 30-minute intervals for all active sectors. A Principal Components Analysis (PCA) was conducted to evaluate the current set of POWER variables and provide guidelines for the addition of new measures or the modification of existing ones. PCA with Varimax rotation converged in seven iterations and produced five components with eigenvalues > 1. Cumulatively, the four components accounted for 68.18% of the variability in the data set: Component 1 (Activity) accounted for 26%, Component 2 (Flight Path Variability) accounted for nearly 13%, Component 3 (Objective Workload) accounted for 11%, Component 4 (D-side Activity) accounted for 9%, and Component 5 (Overload) accounted for approximately 8%. Variables comprising the five extracted components provided valuable information about the underlying dimensions of the NAS data set. Additions or modifications that might improve the ability of POWER to describe ATC activity and taskload were identified.17.
This work was accomplished under the approved task AAM-A-HRR-516.
AbstractThirty full performance level (FPL) en route air traffic control specialists participated in an investigation of the salient features of aircraft mix, a proposed sector complexity factor. Controllers rated the "familiarity" (i.e., frequency of encounter) of 30 selected aircraft. They also provided weight class, engine number, engine type, cruising speed, climb, and descent rate estimates for each aircraft. A matrix of squared Euclidean distances derived from summary estimates (i.e., means of speed, climb, and descent) was used to construct a multidimensional scaling model of the aircraft. Multiple regression interpretation revealed that Dimension 1 was related to engine type, whereas Dimension 2 was associated with weight class. The position of elements in the derived stimulus space indicated that controllers may develop performancerelated prototypes through the use of multiple cues derived from a number of sources. Results are presented as justification for further investigation into potential advantages of providing enhanced prediction cues (e.g., engine type and weight class) from a single source, which may increase the efficiency of controller decision making and decrease perceived workload.17.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.