Dynamic velocity scaling for industrial collaborative robots: a gaze-driven approach

Implementing an effective human-robot seamless interaction (HRSI) is a critical challenge for the industrial robotics community, which requires capturing all human mental processes and turn them into suitable robot's actions. This work presents a method to adjust the robot behavior in real-time considering both human's low-frequency, i.e. cognitive workload, and high-frequency cognitive processes, i.e. visual attention, which are generally overlooked in literature, to optimize safety, productivity and ergonomics. Regarding high-frequency processes, the system monitors the operator's gaze - the closer they look to the robot, the more attentive they are, while attention drops when the robot leaves their field of view, posing potential risks. The speed of the manipulator is then dynamically modulated based on operator's visual attention. Regarding low-frequency processes, the robot's trajectory is adjusted to optimize operator's cognitive workload. An experimental validation involving 26 participants led to three key findings: the developed algorithm improved productivity (18% improvement), cognitive workload (5%), fluency (10%), usability (5%), reliability (5%), and acceptance of the system (9%); the exploitation of high-frequency cognitive processes leads to a significant improvement in all mentioned metrics, suggesting its relevance for future research in the field of HRSI; increasing the level of adaptability of the robotic system positively affects collaboration's effectiveness.