Reconstruction of Human Skills by Using PCA and Transferring them to a Robot

Abstract

This paper discusses human skills enabling rapid adaptation to a changing environment, e.g., when a human table tennis player hits an incoming ball, and describes how to transfer these skills to a robot. Human skills are classified into motor and cognitive. Motor skills are functions involving precise limb movement with the intent to perform a specific act, i.e., hitting a ball. Cognitive skills are functions involving meaningful responses to external stimuli. We extract these skills from observing human movement using principal component analysis and generalize these skills as a schema for a generalized motor program. We also describe table tennis matches between a human opponent and a robot to which these skills have been transferred.