2A1-C04 Using Biofeedback to Analyze Human-Robot Interaction Experiments(Cooperation between Human and Machine)

Abstract

In this paper, we discuss methods for using sensor data, combined with biographical and behavioral data, to investigate people's impressions towards a robot during an interaction. We outline a methodology for how to create a relevant and reliable data set from raw sensor data taken from the robot, and how these data can be interpreted psychologically. We conducted a large-scale human-robot interaction experiment at the University of Tokyo campus, where the humanoid robot HRP-2 greeted subjects by bowing, waving, and shaking hands with them. During the interaction, we collected sensor data including temperature, force, and tactile data. We found significant differences between different subject groupings, indicating that biofeedback can be a useful too to implement within a robot's design.