Cognitive workload and complexity in air traffic controlsimulation of operational tasks in eye tracking studies. (carga mental y complejidad en control de tráfico aéreo: simulación de situaciones operativas en estudios con registro de movimientos oculares)

Zusammenfassung

The aviation transportation system has seen an impressive growth in the second half of last century, and the global quantity of air traffic has doubled once every 15 years from 1977. This trend will continue in the next future, as made clear by the International Civil Aviation Organization (ICAO) in forecasts for future air traffic (ICAO, 2013). Controller workload ¿is likely to remain the single greatest functional limitation on the capacity of the ATM system¿ (Majumdar & Polak, 2001). Experienced workload is commonly intended in ATC as mostly determined by cognitive complexity (i.e. the perceived complexity of traffic situation). In fact, ATC tasks and strategies available to order air traffic safely are mostly based on mental activities as monitoring, evaluating, planning, and projecting (Pawlak et al., 1996). The need for objective and unobtrusive cognitive workload measures is today fundamental for ATC system safety. Eye tracking technologies allow recording and processing of cognitively-related visual behavior data with minimal interference on task performance. The study of controllers¿ visual behavior by means of eye metrics recording might further benefit the modeling of ATC workload. Applications of ocular metrics recording in ATC domain are relatively recent, in particular in relation to conflict detection task (Hunter & Parush, 2009; Martin, Cegarra, & Averty, 2011; Kang & Landry, 2010; 2014). Conflict detection is a key component of ATC (Kallus et al., 1999) and one of the most investigated ATC tasks. Cognitive workload has been demonstrated to be affected by predicted conflicting trajectories (Neal & Kwantes, 2009). This Doctoral Thesis presents four experimental studies in which several ocular metrics were measured as workload indexes in simulated ATC conflict detection tasks. Task demands varied by manipulating acknowledged ATC complexity factors. Workload was assessed multidimensionally, as performance metrics and subjective ratings were also recorded. In the first experiment a simple perceptual task about target position was used as easy experimental condition, and additional tasks (paper-and-pencil task and math operation) were added to create a difficult condition. Workload and fatigue were measured as a function of task complexity (TC) and time on task (TOT), by means of eye movements¿ main sequence parameters (saccadic amplitude, peak velocity, and duration), response time, and subjective scales. As expected, TC affected both performance and subjective workload, and TOT affected fatigue scores. Differently, performance improved through time in the higher complexity condition, exclusively. A better processing strategy emerged through time in this condition, while with lower complexity further performance improvements were less possible. Saccadic peak velocity of large saccades (>11°) decreased significantly only in low complexity condition, especially in the first part of experiment. Results suggested that lower complexity condition was excessively simple and peak velocity showed sensitivity to such underload condition. In the second experiment, convergence angle and distance to convergence point were manipulated in a relative judgment task (estimation of arrival order at convergence point). Traffic scenarios were built using an ATC flight traffic simulator. Results confirmed a slow down effect of peak velocity of large saccades (>15°) with wider convergence angles and different (compared to same) distance to convergence point. Saccadic durations showed the same pattern. Therefore, there was probably an increase in precision of large saccades¿ performance with higher geometry complexity. Peak saccadic velocity decreased with higher experienced workload. In the third experiment, dynamic scenarios were used for conflict detection task. Convergence angle (CA) and minimum distance at closest approach (MD) were manipulated as complexity factors. Conflict and no-conflict scenarios were built by implementing MD values below and above safe separation standard (5nm, nautical miles), respectively. Conflict detection showed higher error-proneness when MD was similar to separation standard (e.g. 6nm). However, conflicts were the most demanding scenarios, as showed by convergent workload measures. Perpendicular routes further contributed to experienced workload. Large saccades showed lower peak velocity values for conflicts, especially with right angles of convergence. Fixation data showed no effects once controlled for the effect of time needed to respond, confirming to be informative of processing time and effort. Two different cognitive strategies were probably adopted in conflicts and no conflicts. The differences in speed and distance to convergence point at the beginning of each scenario might have allowed the application of a ratio-based strategy in no conflicts, while requiring a more demanding motion prediction strategy in conflicts. The study suggested that complexity of conflict judgment depends on relative judgments and estimations of speed, distance, and altitude differential values. In the last experiment, differential values of speed, altitude, and distance to convergence point were manipulated as complexity factors. Pupil size was measured as workload index. Gaze transitions between task relevant areas of interest were analyzed in order to infer adopted strategies and assess related workload. Important results emerged from gaze transition analysis. For example, transitions between aircraft data tags (altitude and speed information) were almost absent when lateral separation estimation was more demanding. Furthermore, there was an increase of transitions between aircrafts and convergence center when processing of differential distance was more demanding. Lastly, frequency of crossed transitions between aircraft elements (position and data tags) increased when both lateral and vertical separation estimation were performed. Significant pupil dilations were observed in most demanding situations. However, the greatest pupil dilations were recorded for the easiest scenarios, and they were probably related to an emotional positive reaction to the perceived easiness of decision. Proposed analysis of gaze transitions in ATC conflict detection represents a relatively novel contribution to the study of complexity and workload. The application of eye tracking methods in ATC domain could benefit in the future both the system design (e.g. conflict alert tools) and ATC training, for instance by identifying experience-related differences in visual behavior, or the visual correlates of scanning strategies. Eye tracking methods might be successfully included as additional source of information in future programs for workload assessment. REFERENCES Hunter, A. C., & Parush, A. (2009, October). Using eye movements to uncover conflict detection strategies. In Proceedings of the Human Factors and Ergonomics Society Annual Meeting (Vol. 53, No. 22, pp. 1729-1733). Sage Publications. ICAO, International Civil Aviation Organization, (2013). Global air navigation plan 2013-2028. Doc 9750-AN/963, Fourth Edition, 2013. Kallus, K. W., Van Damme, D., & Dittmann, A. (1999). Integrated task and job analysis of air traffic controllers¿phase 2: task analysis of en-route controllers. Deliverable HUM.ET1.ST01.1000-DEL04. Eurocontrol: Bruxelles. Kang, Z., & Landry, S. J. (2010). Capturing and analyzing visual groupings of multiple moving targets in an aircraft conflict detection task using eye movements. In¿Proceedings of the Human Factors and Ergonomics Society Annual Meeting (Vol. 54, No. 23, pp. 1906-1910). SAGE Publications. Kang, Z., & Landry, S. J. (2014). Using scanpaths as a learning method for a conflict detection task of multiple target tracking. Human Factors, 56(6), 1150-1162. Majumdar, A., & Polak, J. (2001). Estimating the capacity of Europe¿s airspace using a simulation model of air traffic controller workload. Paper No. 01-3250. In 80th Annual Meeting of the Transportation Research Board. Washington, DC: Transportation Research Board. Martin, C., Cegarra, J., & Averty, P. (2011). Analysis of mental workload during en-route air traffic control task execution based on eye-tracking technique. In Engineering Psychology and Cognitive Ergonomics (pp. 592-597). Springer Berlin Heidelberg. Neal, A., & Kwantes, P. J. (2009). An evidence accumulation model for conflict detection performance in a simulated air traffic control task. Human Factors: The Journal of the Human Factors and Ergonomics Society, 51(2), 164-180. Pawlak, W. S., Brinton, C. R., Crouch, K., & Lancaster, K. M. (1996, July). A framework for the evaluation of air traffic control complexity. In Proceedings of the AIAA Guidance Navigation and Control Conference, San Diego, CA.