Basic Research on Laborer State Prediction Towards the Realization of Human Digital Twin

In the manufacturing field, the diversification of consumer needs is forcing a shift to variable-mix, variable-volume production, making it increasingly difficult to achieve the conventional uniformity of machine-centered production, the ability to respond to diverse needs, and the improvement of production efficiency. In addition, Japan will become a super-aging society where 40% of the population will be elderly by 2060, and there is an urgent need to restructure the securing and utilization of human resources. To solve these issues, a new manufacturing system in which “people” play a leading role has been proposed. To realize this system, the human-digital twin is attracting attention. The human-digital twin is the reproduction in digital space of an individual’s physical, behavioral, and psychological states in the real world, which is thought to enable prediction of laborer fatigue, improvement of work efficiency, and enhancement of laborer safety. This study conducts basic research on predicting laborer fatigue toward the realization of the human digital twin. By conducting demonstration experiments assuming a cell production site, acquiring biometric information, and analyzing it using an autoencoder, it is clarified the prediction of laborer fatigue and the relationship between biometric information and fatigue.