Disturbance Preview-Based Output Feedback Predictive Control for Pneumatic Artificial Muscle Robot Systems With Hysteresis Compensation

Pneumatic artificial muscles (PAMs) exhibit various advantages in human–robot interactions, such as excellent flexibility, high power-to-weight ratios, lightweight materials, and so on; however, some inherent characteristics of PAMs, e.g., complex hysteresis nonlinearities, saturation, and input constraints, may increase control difficulties and deteriorate positioning/tracking performance. Then, multiple working environments unavoidably introduce uncertainties and disturbances to PAM robot systems. In this article, a
new
robust output feedback predictive control method is proposed for PAM robot systems, and hysteresis compensation including initial loading curves is introduced to transform the complicated nonlinear system into a concise linear system
instead of
implementing linearization operations. Moreover, discrete-time high-order sliding-mode differentiators are utilized to estimate lumped disturbances and their high-order derivatives, which are accurately considered to obtain high-precision model prediction. In particular, by utilizing the hysteresis compensation, this article proposes the
first
solution to realize model simplification of PAMs, which
significantly
reduces computation costs and improves control efficiency. Finally, various experimental results on self-built single PAM robot and 2-DOF delta PAM robot platforms are provided to validate the effectiveness and feasibility of the presented method.