Adaptive embodied entrainment control based on communication activity measurement - A challenge for a robotic introducer agent -

In this paper, we propose adaptive control of robotic embodied entrainment to multi communication participants. That is based on communication activity measurement, and uses interaction timing learning which depends on nonverbal communication channels. Our mechanism selects an appropriate embodied robotic behavior, and increases the communication activity by changes of a speaker and a listener. For this, we focus on a gaze lead, gaze distribution and a synchronizing nod. The actions and timing are controlled by a decision-tree. We applied our proposal to a robotic introducer agent and experiment. As the result, the agent could control communication situations similarly to a human.