Closed-loop Control of a Pneumatic Artificial Muscle Actuated 2-DOF Delta Mechanism with Adaptive Hysteresis Compensation

Abstract

As a flexible bionic actuator, pneumatic artificial muscle (PAM) has been widely utilized in robotics. However, due to its strong creep and hysteresis nonlinearities, it is difficult to precisely control its output force and displacement. In order to attenuate the influence of the PAM's hysteresis, this paper proposes the integration of the inverse Prandtl-Ishlinskii (PI) model and the adaptive projection (AP) algorithm such that the model parameters can be identified online. This helps to compensate for the PAM's hysteresis without any offline identification. Finally, the performance of the proposed method is experimentally verified on an in-house built 2-DOF Delta mechanism. Both the angle tracking performance of the joints and the trajectory tracking performance of the end effector are tested. Experimental results demonstrate the effectiveness of the proposed controller.