Assessing operator stress in collaborative robotics: A multimodal approach
本研究通过记录操作员在协作机器人编程任务中的脑电图、心电图、皮电反应和面部表情等多模态生理信号，结合主观问卷，筛选出与主观压力感知最一致的生理参数，为实时监测操作员压力提供依据。
In the era of Industry 4.0, the study of Human–Robot Collaboration (HRC) in advancing modern manufacturing and automation is paramount. An operator approaching a collaborative robot (cobot) may have feelings of distrust, and experience discomfort and stress, especially during the early stages of training. Human factors cannot be neglected: for efficient implementation, the complex psycho-physiological state and responses of the operator must be taken into consideration. In this study, volunteers were asked to carry out a set of cobot programming tasks, while several physiological signals, such as electroencephalogram (EEG), electrocardiogram (ECG), Galvanic skin response (GSR), and facial expressions were recorded. In addition, a subjective questionnaire (NASA-TLX) was administered at the end, to assess if the derived physiological parameters are related to the subjective perception of stress. Parameters exhibiting a higher degree of alignment with subjective perception are mean Theta (76.67%), Alpha (70.53%) and Beta (67.65%) power extracted from EEG, recovery time (72.86%) and rise time (71.43%) extracted from GSR and heart rate variability (HRV) metrics PNN25 (71.58%), SDNN (70.53%), PNN50 (68.95%) and RMSSD (66.84%). Parameters extracted from raw RR Intervals appear to be more variable and less accurate (42.11%) so as recorded emotions (51.43%). • Real-time monitoring of operator stress in Industry 4.0 is essential. • Different psychophysiological signals recorded in real-time with modern biosensors. • The subjective stress response is very variable and dependent on various factors. • Record if variations in physiological parameters agree with individual perceptions. • Select metrics best in accordance with subjective stress perception.