Assessing mental fatigue in mobile crane operators using non-invasive wearable sensor: A study on construction site

Cranes are crucial heavy construction equipment and are among the most dangerous machines on construction sites. This study introduces a novel approach that implements non-invasive wearable technology in a live construction site environment rather than a laboratory simulation environment, which does not reflect a real site environment. The paper provides a new understanding of how digital technologies and data-driven methods can support engineering processes and decision-making for assessing workers’ fatigue in a live construction site environment. The study leverages objective data streams (e.g., BVP, EDA, and ST) to train predictive models that can classify fatigue levels with high accuracy. It demonstrates the role of engineering informatics in augmenting human–machine systems, marking a dynamic step toward the development of adaptive, intelligent construction environments where engineering knowledge is systematically embedded in real-time. The physiological signals from five crane driver volunteers were recorded to assess their mental fatigue during real working conditions involving the loading and unloading of heavy materials at a construction site. Two machine learning models were utilized: k-Nearest Neighbors (k-NN) and Gradient Boosted Trees (GBT). The k-NN model achieved the highest accuracy of 90.74% for the two-class classification and 85.49% for the three-class classification. These findings contribute to engineering informatics by introducing a novel insight into how computational methods can formalize engineering knowledge in the domain of Heavy construction machinery safety which marks a significant step towards automated safety measures for decision makers.