Style based robotic motion

In this paper, we present an approach to motion sequencing and generation in which “style of motion” is taken into account in a systematic manner. In particular, we present a method for injecting so-called dynamic efforts into a discrete motion sequencing framework that utilizes existing theory of stylistic human movement to inform a principled approach to generating trajectories. Namely, choosing weights in a linear-quadratic cost function leads to trajectories corresponding to the eight basic effort qualities found in dance theory; each weight scales a different motion factor that describes an element of style perceived by an audience. Combined with a style-sensitive motion sequencing scheme, we can fully describe stylistic system behavior. Thus, this paper also reports on the application of this general framework to the problem of generating motion sequences for a humanoid robot that exhibit distinctly different stylistic behaviors though they are composed from the same underlying building blocks.