Physiological workload assessment for highly flexible fine-motory assembly tasks using machine learning

In assembly of small-volume products, tasks are still frequently executed manually. However, the lead times foreseen for these tasks, often do not take into account the actual capabilities of the employees, which in turn leads to increased workload and the associated stress among the employees. This paper investigates how a commercially available wearable low-cost sensor and two machine learning algorithms can be applied to measure and evaluate heart rate, heart rate variability and respiration rate to establish a relationship with workload. The investigated algorithms, namely Random Forest and K-Nearest-Neighbours are able to distinguish between tasks phases and rest phases as well as between easy and difficult tasks executed by the employee, which is the main novelty of this paper.