Control Degradation Outperforms Physiological Measures in Indicating Workload Levels During Robotic Teleoperation
Robotics holds the potential to streamline the execution of repetitive and dangerous tasks, which are difficult or impossible for a human operator. However, in complex scenarios, such as nuclear waste management or disaster response, full automation often proves unfeasible due to the diverse and intricate nature of tasks, coupled with unpredictable hazards, and is typically prevented by stringent regulatory frameworks. Consequently, the predominant approach to managing activities in such settings remains human teleoperation. However, teleoperation can be demanding, especially in high-stress situations, and involves a complex blend of both mental and physical. We present an experiment to explore a range of physiological and performance-related metrics for workload assessment during robotic teleoperation. 35 participants performed a teleoperation task during which we manipulated cognitive and physical workload conditions. We recorded multiple metrics including brain activity using functional Near-Infrared Spectroscopy (fNIRS), galvanic skin response (GSR), heart rate (HR) and variability (HRV), subjective workload ratings, task performance, and robot kinematics. Our results suggest that robotic control degradation may be the most robust metric for distinguishing between different levels of workload experienced during teleoperation, with most physiological differences in the level of mental workload becoming insignificant during a high-workload task.