<XML><RECORDS><RECORD><REFERENCE_TYPE>3</REFERENCE_TYPE><REFNUM>7362</REFNUM><AUTHORS><AUTHOR>Hale,J.G.</AUTHOR><AUTHOR>Pollick,F.E.</AUTHOR></AUTHORS><YEAR>2000</YEAR><TITLE>Sticky Hands interaction with an anthropomorphic robot</TITLE><PLACE_PUBLISHED>Workshop on Interactive Robotics and Entertainment (WIRE-2000), Carnegie Mellon University, Pittsburgh, Pennsylvania </PLACE_PUBLISHED><PUBLISHER>Dept of Computing Science, University of Glasgow</PUBLISHER><LABEL>Hale:2000:7362</LABEL><ABSTRACT>Sticky hands is an exercise that involves two players placing their hands in contact and moving the contact point while attempting to maintain a constant minimal contact force. It requires players to be sensitive and responsive to their partner s motions. It is used to develop coordination, relaxation, grace and is a comfortably intimate interaction between human beings. We present a system for playing with an anthropomorphic robot intended to explore robotics use in self-development and to facilitate intimate human-like interactions with the robot. We discuss motor control of our robot briefly, then present a new learning algorithm capable of generalizing observed motions to the performance of similar movements. Our algorithm records and maintains a set of prototypes that describe hand motions using instantaneous properties of the hand s trajectory. We discuss methods of simulating human-like, emotional and expert sticky hands styles of movement drawing upon the fields of human motor production and perception, and computer graphics. </ABSTRACT></RECORD></RECORDS></XML>