Exploring Engagement in Human-Robot Interaction through the Quantification of Human Spontaneous Movement
Résumé
Developments in the field of social robotics open
interesting opportunities for applications in healthcare, education,
and services. For this, studying engagement in humanrobot
interaction (HRI) is crucial for improving the quality
of interactive experiences. Questionnaires are powerful in describing
voluntary behavior; however, engagement is often an
implicit non-voluntary behavior that reaches awareness only
once initiated. Inspired by research in cognitive psychology,
we propose a behavioral feature to quantify engagement in
HRI through the measurement of spontaneous movement and
spectral wavelet analysis. For this, we conducted an experiment
during which participants listened to sad stories narrated by a
moving social robot. Throughout the experiment, we tracked the
participants’ spontaneous and non-voluntary sway movements
with a motion capture system. The experiments were conducted
with three robotic platforms (Buddy, Pepper, and Nao). Results
showed that spontaneous body sway can be modulated by social
robots within no-goal-oriented interaction. This opens up the
application possibility of our methodology to various modes of
HRI such as collaboration.