A Field Study of Work Type Influence on Air Traffic Controllers’ Fatigue Based on Data-Driven PERCLOS Detection

Abstract

The fatigue of air traffic controllers (ATCOs) on duty seriously threatens air traffic safety and needs to be managed. ATCOs perform several different types of work, with each type of work having different characteristics. Nonetheless, the influence of work type on an ATCO’s fatigue has yet to be demonstrated. Here, we present a field study in which the fatigue of ATCOs working in two types of work was compared based on an optimized data-driven method that was employed to detect the percentage of eyelid closure over the pupil over time (PERCLOS). Sixty-seven ATCOs working within two typical jobs (i.e., from the terminal control unit (TCU) and area control unit (ACU)) were recruited, and their fatigue was detected immediately before and after shift work using PERCLOS. Using a Spearman correlation test analysis, the results showed that the influence of work type on an ATCO’s fatigue had interesting trends. Specifically, the ATCOs at the TCU who handle departures and arrivals, which include converging with and maneuvering around conflicts, retain normal circadian rhythms. Their fatigue was significantly influenced by the various demands from tasks focusing on sequencing and conflict resolution and by the time phase of a normal circadian rhythm. At the ACU, ATCOs manage flights that are mainly on route, causing monotonous monitoring and routine reporting tasks, and the ATCOs generally have frequent night shifts to handle overflights. Their fatigue was significantly influenced by the demand characteristics from tasks, but changes in fatigue rule were not consistent with a normal circadian rhythm, revealing that the ATCOs’ circadian rhythms may have already been slightly disturbed. Furthermore, the interactions between task demand and circadian rhythm with an ATCO’s fatigue were significantly observed in ATCOs working in the TCU but not in those in the ACU. This study provides first evidence that an ATCO’s work type influences his or her fatigue. This discovery may incite stakeholders to consider work type in the management of employee fatigue, not only in the civil aviation industry but also in other transport industries.

PERCLOS-based technologies for detecting drowsiness: current evidence and future directions

“…Previous studies have assumed complete eye closure at a given instant to be 0%, with (1) 100% representing eyes wide open and 0% representing eyes closed [ 45 ]; (2) the percentage of the distance between eyelids when the diameter of the iris is represented as 100% [ 47 , 51 ]; (3) the percentage of the pupil occluded by the eyelid with the diameter of the pupil was 100% [ 52 ]. Eye-opening of less than 25% [ 53 ] or 30% [ 47 , 54–57 ] has also been used as a definition of eye closure in some previous studies. In some cases, to obtain a measure of PERCLOS in a simplified manner, the eye is considered to be closed (4) when the pupil cannot be detected [ 58 , 59 ], or (5) the eye aspect ratio [ 60 ] is below a definite value [ 61 ].…”

“…The definitions of fast eyelid closure include <250 ms [ 47 , 64 ], <400 ms [ 52 , 62 ], or <500 ms [ 51 ]. The sampling frequencies for recording the degrees of eyelid opening include 2 Hz [ 65 ], 3 Hz [ 66 ], 6 Hz [ 56 ], 10 Hz [ 51 , 67 ], 24 Hz [ 54 ], 60 Hz [ 68 , 69 ], or 120 Hz [ 52 , 55 ]. A sufficient sampling rate is required when fast blinks are excluded.…”

A fatigue driving detection algorithm based YOLOv5

“…PERCLOS [17] is the most authoritative physical quantity to measure fatigue proposed by foreign researches. The Federal Highway Administration has gathered a large number of scholars to conduct intensive research and comparison on fatigue driving detection indicators, and explore the most effective indicators that can represent fatigue.…”

Machine Learning for Fatigue Detection using Fitbit Fitness Trackers

“…An alternative method is percentage eye openness tracking (PERCLOS) (Zhang et al, 2021). PERC-LOS tools rely on continuous video capturing of a person's sight and using image processing in order to determine where the eyes are located, and whether current eye movements can be correlated to the state of fatigue.…”