Measurement of involvement in human-robot interaction produced by eye contact

Abstract

— The present study aimed at investigating how eyecontact established by a humanoid robot affects engagement inhuman-robot interaction (HRI). To this end, we combinedexplicit subjective evaluations with implicit measures, iereaction times and eye tracking. More specifically, we employeda gaze cueing paradigm in HRI protocol involving the iCubrobot. Critically, before moving its gaze, iCub eitherestablished eye contact or not with the user. We investigatedthe patterns of fixations of participants' gaze on the robot'sface, joint attention and the subjective ratings of engagement asa function of eye contact or no eye contact. We found that eyecontact affected implicit measures of engagement, ie longerfixation times on the robot's face during eye contact, and jointattention elicited only after the robot established eye contact.On the contrary, explicit measures of engagement with therobot did not vary across conditions. Our results highlight thevalue of combining explicit with implicit measures in an HRIprotocol in order to unveil underlying human cognitivemechanisms, which might be at stake during the interactions.These mechanisms could be crucial for establishing an effectiveand engaging HRI, and could potentially provide guidelines tothe robotics community with respect to better robot design.