Joint stiffness and position control of an artificial muscle manipulator considering instantaneous load

Abstract

In recent years, robots have been used in medicine and everyday life. Therefore, it is desirable that these robots be flexible and lightweight. For this reason, we have studied and developed straight-fiber-type artificial muscles derived from McKibben-type muscles, which have excellent contraction rate and force characteristics. Last year, we developed a manipulator with six degrees of freedom using artificial muscles as actuators, and have considered its position control. However, artificial muscle manipulators are susceptible to load torque because they do not use any gears and are flexible. Joint stiffness must increase because accurate position control of the artificial muscle is difficult. The stiffness control must respond quickly to achieve coordination with human activity. However, conventional stiffness control by torque-based methods depends on the position control response. Therefore, position control and stiffness control need to be independent. In this study, we propose a new method of joint stiffness control, which adds estimated stiffness to the torque-based method. In addition, we performed experiments examining load response to the steady state and instantaneous loads.