“Synchrony” as a way to choose an interacting partner

Robots are poised to fill a growing number of roles in todays society. In future, we could have robots expected to behave as companion in our home, offices etc. Moving to social robotics imply to address several issues related to human-robot interactions for instance, how the robot can develop an attentional mechanism and select an interacting agent among several interactants. We took our inspiration from neurobiological and psychological studies suggesting that synchrony is an essential parameter for social interaction. We assumed that synchrony detection could be used for intitiating human robot interaction. We present a neural network architecture able to focus the attention of the robot and to select an interacting partner on the basis of synchrony detection.