Task Space Approach of Robust Nonlinear Control for a 6 DOF Parallel Manipulator

Abstract

The dynamics and kinematics of a parallel manipulator has been widely researched by virtue of its a high force-to-weight ratio and widespread applications ranging from vehicle or flight simulator to machine tool despite a smaller workspace than a serial robot system (Merlet, 2000). Such a parallel system has been paid special attention as a typical multi-input multi-output nonlinear system to retain a high control performance. A control scheme for a 6 DOF parallel manipulator can be classified into two groups: a joint space based control scheme (Honegger et al., 2000; Kang et al., 1996; Kim et al., 2000; Nguyen et al., 1993; Sirouspour & Salcudean, 2001) and a task space based control strategy (Kang et al., 1996; Park, 1999; Ting et al., 1999). It is easy to realize the joint space based control scheme to a parallel manipulator as if the decoupled single-input single-output (SISO) control systems activate for a parallel mechanism. The simplicity has let many research activities pursue more specific approaches. As a result, the novel joint space based control approaches have been studied to improve the control performance by rejecting the nonlinear effects in the equations of motion (Honegger et al., 2000; Kang et al., 1996; Kim et al., 2000; Nguyen et al., 1993; Sirouspour & Salcudean, 2001). Particularly, for a parallel system driven by a hydraulic-servo system, joint space based robust nonlinear control scheme (Kim et al., 2000) has proposed. However, the research has dealt with excessively conservative uncertainties including gravity and known dynamic characteristics even though the friction effect can be neglected by the hydrostatic bearing. On the other hand, a task space based control for a 6 DOF parallel manipulator has a potential to meet excellent control performances under system uncertainties: inertia, modeling error, friction, etc. However, its scheme may be realized by the obtained the 6 DOF system state through a costly sensor or a novel nonlinear state estimation methodology. H∞ robust control strategy (Park, 1999) and the adaptive control scheme (Ting et al., 1999) have been studied as the examples of task space based control. However, there have been still some weak points in the previous researches; the linearized model based approach and a simulation study only, respectively. Another task space based nonlinear control scheme has been proposed to a Stewart platform (Kang et al., 1996). However, it has also shown the computational simulation results only on the assumption to the system uncertainties that seems excessive. Furthermore, its treatment on stick-slip friction is minimal, which may give rise to serious deterioration of control performance in a real system where the frictional property is not negligible.