Identification of Gracefulness Feature Parameters for Hand-Over Motion

摘要

As robots in the welfare and service industries must come into contact with humans, they are required to make favorable impressions on human sensibilities. Our research focuses on the concept of "graceful motion," as defined by Hogarth. In order to implement graceful motion in robots, we analyze the hand trajectories which highly skilled servers generate in the task of passing a wine glass to extract graceful and ungraceful curve features. We propose to model the hand trajectory by a polynomial of the fourth degree and to adjust the S-shaped curvature of the hand trajectory. An impression evaluation is then conducted, which indicates that a 20 % - 60 % S-shaped curvature correspond to gracefulness in a hand-over motion; this parameter corresponds to Hogarth's definition.