Identification and classification of construction equipment operators' mental fatigue using wearable eye-tracking technology

• Published in: Automation in construction Vol. 109; p. 103000
• Main Authors: Li, Jue, Li, Heng, Umer, Waleed, Wang, Hongwei, Xing, Xuejiao, Zhao, Shukai, Hou, Jun
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01.01.2020; Elsevier BV
• Subjects: Accidents; Algorithms; Classification; Clustering; Computer simulation; Construction equipment; Construction equipment operator; Construction industry; Covariance; Excavators; Eye movements; Eye-tracking; Fatigue; Feasibility studies; Feature extraction; Machine learning; Mental fatigue identification and classification; Operators (personnel); Support vector machines; Technology utilization; Toeplitz Inverse Covariance-Based Clustering; Tracking; Wearable technology; Construction equipment operator; Toeplitz Inverse Covariance-Based Clustering; Eye-tracking; Mental fatigue identification and classification; Machine learning
• DOI: 10.1016/j.autcon.2019.103000
• ISSN: 0926-5805

In the construction industry, the operator's mental fatigue is one of the most important causes of construction equipment-related accidents. Mental fatigue can easily lead to poor performance of construction equipment operations and accidents in the worst case scenario. Hence, it is necessary to propose an objective method that can accurately detect multiple levels of mental fatigue of construction equipment operators. To address such issue, this paper develops a novel method to identify and classify operator's multi-level mental fatigue using wearable eye-tracking technology. For the purpose, six participants were recruited to perform a simulated excavator operation experiment to obtain relevant data. First, a Toeplitz Inverse Covariance-Based Clustering (TICC) method was used to determine the number of levels of mental fatigue using relevant subjective and objective data collected during the experiments. The results revealed the number of mental fatigue levels to be 3 using TICC-based method. Second, four eye movement feature-sets suitable for different construction scenarios were extracted and supervised learning algorithms were used to classify multi-level mental fatigue of the operator. The classification performance analysis of the supervised learning algorithms showed Support Vector Machine (SVM) was the most suitable algorithm to classify mental fatigue in the face of various construction scenarios and subject bias (accuracy between 79.5% and 85.0%). Overall, this study demonstrates the feasibility of applying wearable eye-tracking technology to identify and classify the mental fatigue of construction equipment operators. •A novel approach for detecting construction equipment operator’s mental fatigue based on eye movement data is proposed.•TICC method is adopted to identify multiple levels of mental fatigue and label relevant eye movement data.•Supervised learning algorithms are used to learn four different sets of eye movement features.•Results show that SVM and LDA yield high classification performance.•The feasibility of the proposed method for detecting multi-level mental fatigue is discussed.