Biomimetic actuator control for a pneumatic cylinder

Air driven servo-pneumatic actuators are of common use in many industrial applications. System inherent advantages like robustness, power to weight ratio, speed and clean operation conditions suggest an application to rehabilitation robotics. This paper examines impedance modulation properties of antagonistically working human muscles via an extended version of the the Hill-model and a λ-equilibrium-point controller. A control strategy derived to mimic the impedance behaviour of the human muscle including neural feedback is proposed for a linear pneumatic cylinder. Inner impedance and outer average mean pressure control loops are designed for an example cylinder via an optimal linear-quadratic regulator approach. The control strategy is extended with a friction compensation and tested in different load scenarios. Impedance of the pneumatic cylinder is compared to antagonistically working human muscles of similar maximum force capability in the frequency range.