Evaluating mental workload and affective behavior in human-robot collaboration in industrial context: a neuroscientific multidimensional approach

The introduction of collaborative robots, or cobots, in industry has created new ways of interaction between humans and robots. Cobots are designed to work alongside humans toward a shared goal, combining robots’ strength and precision with human flexibility. Developing "intelligent” cobots that adapt to operators’ mental and physical workload would further enhance human-robot collaboration (HRC). This has led to a growing interest in measuring individuals’ mental workload during collaborative tasks with cobots. Such research has primarily relied on self-report measurements combined with autonomic indices or performance measures. A minority of studies have also explored central brain activity using electroencephalography (EEG) or functional Near-Infrared Spectroscopy (fNIRS), but comprehensive studies that integrate all these metrics are lacking. Furthermore, potential variables impacting mental workload are often overlooked. Given that mental workload is a multifaceted construct influenced by cognitive, emotional, social, and environmental factors, a multidimensional approach to its study is suggested. A research protocol is proposed with the aim of identifying mental workload and emotional biomarkers associated with both central and peripheral activity and evaluating individual characteristics impacting the quality of HRC. Understanding the dynamics of HRC from an integrated perspective, combining neurophysiological, behavioral, and subjective measurements, while considering individual differences and cognitive functions in mental workload assessment, is intended as the initial step towards developing cobots that adapt to humans needs. With this aim, neurophysiological and behavioral metrics of mental workload could be useful in the development of real-time prediction and decoding models of individuals’ psychophysiological states for implementation in cobots. For this purpose, a sample of 40 subjects will undergo two experimental phases. First, cognitive functions and personality will be assessed using neuropsychological tests and questionnaires. EEG, cardiovascular and electrodermal activity will be recorded while performing cognitive tasks and a simple assembling task, establishing a baseline for the cognitive functioning of individuals, and mapping which mental functions are involved in different phases of the assembling task. In the second phase, participants will perform assembling tasks with and without cobot assistance. In addition, relevant factors related to cobot, environment and human-cobot interaction (e.g., speed, proximity, number of exchanges) will be manipulated through different experimental conditions. During tasks performance, a co-registration of brain activity using EEG and fNIRS will be conducted. Cardiovascular and electrodermal activity will also be recorded, and eye movements will be registered using an eye-tracking device. Finally, task performance and self-report measurements will be collected. Differences in EEG frequency bands and in levels of oxygenation of the prefrontal cortex will allow exploration of which conditions are most mentally demanding. Increases in cardiovascular and electrodermal activity and analysis of eye movement patterns will provide additional measures of changes in individuals’ fatigue, mental stress, and attention within the different conditions. These implicit measures of mental workload will then be compared with the objective performance and the subjective experience of participants. Finally, we expect that relevant individual characteristics, such as personality, executive functioning, anxiety, empathy, motivational tendencies, and expectations towards robot and technology, will impact the neurophysiological and subjective response to HRC.