Motion generation for the stuffed-toy robot

A number of entertainment robots to entertain or heal people have applied hard mechanism in order to equip sensors for taking surrounding conditions, or making various motions. However, in point of familiarity, softness and good hand feeling are important factors for the robots. A stuffed-toy robot we proposed has five motion parts made of fabrics (cloth, cotton and thread) and can various body motions. In this paper, we propose a reaction generation method for the stuffed-toy robot. In this method, we use external sensors to get positions and actions of users around the robot. And we derive the amount of attention of robot to each object based on these data. Thereby, the robot gazes and reaches to a target decided the amount of attention. It is expected that the robot is able to react to varied user inputs.