Human skill analysis based on multisensory data

Abstract

In this paper, we report our approach to determine the essential features of human physical skills and the possibility for transferring them to robots. We focus on two tasks: crank rotation and side matching. The human skills are experimentally analyzed by measuring the hand position and force in the execution of these tasks using a vision tracking system and six-axis force/torque sensors. In the experiments, we obtained data under various conditions, such as both using and not using a blinder. Relevant data analysis for characterizing the human skills is presented.