OCCUPATIONAL APPLICATIONSAn understanding of fluency in human-robot teaming from a physiological standpoint is still incomplete. In our experimental study involving 24 participants, we designed a scenario for shared-space human-robot collaboration (HRC) for a material sorting task. When compared to a sequential mode of interaction, the simultaneous mode resulted in significantly higher perceptions of fluency and engagement, primarily by reducing human idle time. These observations were complemented by significant changes in physiological responses, such as ECG entropy and low frequency power. These responses could predict fluency and engagement with accuracies of 90 and 97%, respectively. Notably, the perception of fluency and preferred mode of interaction were influenced by individual preferences. Hence, it is crucial to consider both physiological responses and user preferences when designing HRC systems, to ensure a positive experience with the robot teammate and to foster engagement in long-term teamwork. Furthermore, these signals can be obtained using a single robust, low-cost, and comfortable sensor.
Keywords: ECG; Human-robot collaboration; engagement; fluency; heart rate variability; industrial robots.
Background In the current industry, a key enabler of flexible manufacturing is human-robot collaboration (HRC), a scenario wherein a human and a robot interact and work together in a shared space to accomplish a common task. In HRC, the timing and coordination between the human and robot are crucial factors that impact the fluency, efficacy, and acceptance of human-robot teams.Purpose Experimental research on the physiological indicators of fluency in human-robot collaborative tasks in a shared workspace is still in its infancy. We posit that by relating the mental perceptions of fluency to features from physiological signals, we could bring more clarity to the complex mapping between subjective and objective measures of fluency.Methods Twenty-four participants (12 males and 12 females), with mean (SD) age = 25.7 (2.9) years, completed an experimental study. We investigated the effects of interaction mode (sequential, simultaneous) and level of human involvement (low, medium, high) on perceived fluency, engagement, performance, and physiological response (heart rate variability = HRV) in a collaborative item sorting task.Results The simultaneous mode of interaction and a higher level of human involvement led to higher ratings for fluency and engagement, along with ECG changes, specifically an 11.6% increase in low-frequency power and a 3% reduction in information entropy. Using machine learning, these HRV features could predict perceived fluency and engagement with 90 and 97% accuracy, respectively.Conclusion Our results indicate that a human operator’s perceived fluency in human-robot collaborative tasks can be measured using HRV metrics. Our findings expand the current fluency framework from a physiological perspective and offer additional objective measures derived from HRV, which could be practically applied to improve the design and optimization of HRC systems.