Using Virtual Reality to Monitor Worker Stress and Fatigue in Industry

This study presents a new approach to the training of production line workers in the automotive industry using virtual reality and bio-signals. This approach integrates virtual reality with audio, facial recognition, ECG and EDA signals and the use of machine learning algorithms. The aim of the study was to design a comprehensive system to analyze emotional states such as stress and fatigue and collect selected physiological signals during simulated work on a production line. For this purpose, two scenarios typical of stress and fatigue monitoring on a production line in the automotive industry were proposed. The system uses these signals to semi-automatically classify the test workers into 5 levels and, using machine learning algorithms, assigns them to the appropriate class of desired work activities according to their performance and reactions to stress and fatigue. The quality assessment of the proposed integrated methodology was successfully monitored and tested on a sample of $\mathbf{1 2 0}$ adepts during two virtual reality work activity scenarios. The classification Accuracy of the Random Tree machine learning algorithm was 83% and 81% for stress and fatigue, respectively.