Skill Analysis in Human Tele-operation Using Dynamic Image

Abstract

Most ordinary machines are not designed to assist human to improve one's skill. Human is often demanded to learn an operation of a machine implicitly. To improve this situation, new-generation machine has to consider user's skill, and to assist the human-machine system to achieve best performance. As the first stage, a function to evaluate user's skill is needed to the new mechatronics. In this paper, to obtain knowledge about human skill on machine manipulation, process of a up-skilling is analyzed. In order to obtain experimental data for the skill analysis, we utilizes a tele-operated robot system. For tele-operation, two kinds of skill: machine manipulation skill and environmental cognition skill are needed. For the former, human control characteristics are estimated with an ARX model identification technique, and we investigated correlation with time-delay and control characteristics of the operators. For the latter, ratio of feasible action was analyzed by checking the front camera's view based on image processing analysis. As a results, it could be confirmed that combination of the correlation factor gives classification of the operator's type, and characteristics of the up-skilling process