Study of Smooth and Accurate Position Controls of Pneumatic Artificial Muscle Actuators for Robotic Arms

This paper utilizes position errors, position error rates and payload variations as inner feedback loops for controlling the actuators of joint rotations in a pneumatic artificial muscle (PAM) driven robot. The result is an algorithm that is able to adaptively regulate input signals for the proportional pressure valve that controls the air flow to the PAM, in an attempt to better mimic the point-to-point movement of human arm. Further, the algorithm is able to utilize the angular position information from an outer feedback loop to activate/deactivate a solenoid valve for a rapid damping of the non-oscillatory motion and thereby, able to quickly and accurately bring the joints to a rest at a desired position. The proposed method appears to be effective in overcoming some of the inherent drawbacks that arise from the viscoelastic shell and air compressibility of the PAM element.