Dynamically stable movement generation of a humanoid robot from demonstration: kicking a ball

Abstract

This paper presents a new algorithm of autonomously generating dynamically stable movements such as ball kicking through learning from demonstration by a humanoid robot. It is based on the framework of dynamic movement primitives (DMP) [1, 2] which represent a demonstrated movement with a set of different equations. By suggesting a modification to take into account the dynamic stability condition using the zero moment point (ZMP), the framework is extended to enable to provide dynamically stable movements which reach target positions accordingly. We validate the feasibility of our algorithm through the simulation study and experiment of ball kicking movements by a humanoid robot.