Wearable network for multilevel physical fatigue prediction in manufacturing workers

Area of Science:

• Industrial Engineering
• Biomedical Engineering
• Machine Learning

Background:

• Manufacturing work involves strenuous physical activity leading to fatigue, impacting worker health and incurring significant costs.
• Current methods for fatigue monitoring are often limited, lacking real-time capabilities and a nuanced understanding of fatigue levels.
• Continuous monitoring and feedback are essential for mitigating losses and improving safety in manufacturing environments.

Purpose of the Study:

• To develop and evaluate a novel system for real-time, multilevel physical fatigue quantification in manufacturing workers.
• To address the limitations of dichotomous fatigue assessment by predicting nuanced fatigue states.
• To provide a practical solution for monitoring fatigue on the factory floor using wearable technology.

Main Methods:

• Utilized a multimodal wearable sensor framework to collect vital signs (heart rate, HRV, skin temperature) and locomotive data.
• Employed inertial motion units at six upper-body locations for comprehensive torso and arm movement capture.
• Developed an asymmetric loss function for a machine learning model to enhance prediction accuracy and enable real-time inference.

Main Results:

• Successfully collected data from 43 subjects performing an authentic manufacturing protocol, correlating sensor data with self-reported fatigue.
• Demonstrated the system's ability to predict multilevel fatigue, offering a more detailed assessment than binary methods.
• Validated the system's practical applicability through an in-the-wild evaluation with actual factory operators.

Conclusions:

• The developed system offers a practical and effective solution for continuous, real-time monitoring of physical fatigue in manufacturing settings.
• The novel approach to multilevel fatigue prediction provides deeper insights into worker physical states.
• The study contributes an open-access database, fostering future research in occupational fatigue monitoring.