Identification and Classification of Mind-Wanderings of Drilling Monitoring Personnel Based on Wearable Eye Tracking Technology: A Comparison Under Different Mental Workloads

<span>Objective</span><span>:</span>The study aims to propose an effective method for identifying and classifying mind-wandering states among drilling monitoring personnel. Additionally, it investigates the relationship between mind wandering and mental workload while evaluating the method's performance in categorizing mind-wandering states across various mental workloads. Background: Mind-wandering among drilling personnel can lead to overlooking or misinterpreting critical safety parameters, potentially resulting in serious drilling accidents, hence the necessity for an effective method to identify and classify these states. Method: This study performed drilling monitoring simulations with experienced workers to gather behavioral and eye-tracking data. Hierarchical clustering algorithms were utilized to identify mind-wandering states using the collected data. Supervised learning algorithms were then employed to classify mind-wandering states and evaluate performance. The study also explored the link between mind-wandering and mental workload across various levels, testing the efficacy of the proposed method in identification and classification. Results: Except for Linear Discriminant Analysis (LDA), four supervised learning algorithms exhibited outstanding performance, with all G-Means exceeding 0.95. During mind-wandering, eye-tracking features showed variations across different levels of mental workload. The classification methods achieved an accuracy rate of 84% under various mental workloads, demonstrating significant performance differences based on different combinations of eyetracking features. Conclusion: Proposed an automated method for identifying and classifying mindwandering states using wearable eye-tracking technology, validated across various mental workloads for real-time monitoring and targeted interventions. Application: Implement these research findings in drilling operations to enhance safety by continuously monitoring operators' states in real-time and preventing potential safety hazards.