MIMO controller design for a parallel manipulator system: a practitioner's approach

Abstract

This paper presents a multi-input multi-output (MIMO) controller design procedure for a parallel manipulator system. In particular, a manipulator with a Stewart mechanism is used in the study where the control target is to maintain the orientation of the platform while rejecting disturbances at the base of the manipulator. The investigation starts with a construction of the system model. A direct kinematics equation, an inverse kinematics equation and a closed-loop dynamic equation of the linear actuator are used in the system model formulation. Since the resulting model is a non-linear model, an approximated linear model is subsequently created from the system model using a least square approximation technique. Once the linear system model is available, an MIMO controller is then generated where the feedback controller is designed using a pole placement technique while the disturbance rejection capability is the main aim during the feed-forward controller design. By limiting the operating range of the system, the performance of the linear controller in the closed-loop non-linear system can be estimated via computer simulations. The simulation results indicate that with the use of the designed controller, the effect of sinusoidal disturbances on the system performance can be significantly reduced. The MIMO controller is subsequently tested on a real parallel manipulator developed in-house where some disturbance effects can still be observed from the experimental results. The discrepancies between the simulated and experimental results are believed to be caused by modelling errors.