Operator Engagement Detection for Robot Behavior Adaptation

Abstract

It has been shown that in human-robot interaction, the effectiveness of a robot varies inversely with the operator engagement in the task. Given the importance of maintaining optimal task engagement when working with a robot, it would be immensely useful to have a robotic system that can detect the level of operator engagement and modify its behavior if required. This paper presents a framework for human-robot interaction that allows inference of operator's engagement level through the analysis of his/her physiological signals, and adaptation of robot behavior as a function of the operator's engagement level. Peripheral physiological signals were measured through wearable biofeedback sensors and a control architecture inspired by Riley's original information-flow model was developed to implement such human-robot interaction. The results from affect-elicitation tasks for human participants showed that it was possible to detect engagement through physiological sensing in real-time. An open-loop teleoperation-based robotic experiment was also conducted where the recorded physiological signals were transmitted to the robot in real-time speed to demonstrate that the presented control architecture allowed the robot to adapt its behavior based on operator engagement level.