The McKibben-type artificial muscle has been applied to several industrial fields because of its many advantages over traditional actuators, including, high flexibility, low weight, and high power density. However, the muscle has strong asymmetric hysteresis characteristics that are dependent on the load applied, and these problems prevent its widespread application. This study presents an approach to model hysteresis characteristics of the muscle using an asymmetric Bouc-Wen model. Subsequently, control strategies based on several model predictive control, including servomechanism and adaptive algorithm, and a combination of these methods, were applied to the proposed model. Experimental results revealed that model predictive control with servomechanism showed robustness in terms of load and small variation for control performance. Furthermore, the adaptive model predictive control with servomechanism improved control performance in the steady-state response by 46% compared with the model predictive control with servomechanism and demonstrated robustness in terms of changes in the reference frequency.

A Pneumatic soft actuator has many advantages of a higher ratio of generated force and lightweight and inexpensive. It is also attracting attention because of its intrinsic safety due to compliance based on air compressibility. However, in a system using pneumatic soft actuators, a servo valve is most heavy and expensive. Therefore, in the previous study, a low-cost and small-sized servo valve using buckled tubes was proposed and developed. In this paper, a static analytical model for designing the valve was proposed. In order to decrease the cost and size of valves, a multi-port servo valve that can operate three pneumatic actuators of the flexible robot at the same time was proposed. The design method of the valve based on the analytical model to get desired sinusoidal motion of the robot was described. The driving test of the robot arm using the designed valve based on the model was carried out.

This paper discusses a new uncertainty analysis via low frequency inputs for hydraulic cylinder dynamics. First, we review the nonlinear nominal modeling and the existence of the nonlinearity phenomenon despite the absence of nonlinear friction. Second, we propose a new uncertainty analysis by both physical parameter identification and measurement of the input-output behavior by the identification input within low frequency domain. Finally, we experimentally apply the new uncertainty analysis without ignoring the low frequency region. Unexpectedly, there is the non-negligible uncertainty within low frequency domain which can be important for model-based control of hydraulic robots.