Facilitation of telepresence robot turn-takings by gesture exaggeration

We propose a telepresence robot that exaggerates non-verbal cues for taking turns in teleconferences. In multi-party teleconferences, it is more difficult for the remote participants to take their turns than face-to-face. It is said that an addressee tends to be next speaker. Therefore, becoming the addressee is a previous step to become a speaker and take turns. In order to make the remote participant become addressee, proposed system detected remote participant´s non-verbal cues such as attention directions, nod motions and back-channels. Then, the system exaggerates the cues and expresses them as the robot´s motions.