A Lower Limb Rehabilitation Assistance Training Robot System Driven by an Innovative Pneumatic Artificial Muscle System

This study aims to develop the application of pneumatic artificial muscle (PAM) for a 2-degrees of freedom (2-DOF) lower limb rehabilitation assistance training robot system. The proposed lower limb robot system can be divided into two axes, such as hip joint and knee joint. Each joint contains a pneumatic proportional valve to control a single-PAM with a torsion spring to simulate joint extension and flexion bionics characteristics and achieve a human-like 2-DOF lower limb robot system design and experimental prototype system. By analyzing the kinematics, the derived kinematics conforms to the lower limb motion pattern of the moving human body. Single PAM is difficult to achieve high accuracy control due to the different characteristics between extension and contraction. In our previous research, dual PAMs have been developed to drive a rotational joint which can achieve better control accuracy, however, cannot be suitable for multiaxial robotic design. The mechanism will become very complex and result in lower response and control accuracy. Thus, in this article the novel concept using single PAM with torsion spring was proposed to drive a joint to achieve two-axial robotic design. It has the advantage of multiaxial mechanism design, but the difficulty in joint control due to motion nonlinearity between contraction and extension. The torsion spring can improve motion nonlinearity between contraction and extension partly. Thus, the joint controller using adaptive self-organizing fuzzy sliding mode controller (ASOFSMC) was developed to solve this problem and achieve the required control performance for the joint angle positioning and gait planning control. Through the novel combination of single PAM, torsion spring, and the ASOFSMC joint controller with novel mechanism design and controller design, the two-axial robot mechanism designs and achieves the required control accuracy. The experimental results show that ASOFSMC can effectively control a 2-DOF lower limb robot system, and can modify fuzzy rules online, and adapt to rapid changes in the external environment and load to improve system control performance. The results prove that the proposed innovative single-PAM with a torsion spring and the control strategy can achieve the performance of 2-DOF lower limb rehabilitation assistance training robot system.