Robust unified dual-domain control framework for high-performance parallel robots

This paper introduces a unified joint-task-space control framework for a 6-DoF Stewart platform that overcomes the limitations of pure joint-space methods, including inverse-kinematic ambiguities, configuration flips, and sensitivity to dynamic variations. The proposed architecture integrates a nonsingular fast terminal sliding mode (NFTSM) controller, a nonlinear disturbance observer, and model-based feedforward compensation in the joint space, together with a complementary NFTSM-based task-space controller that continuously refines end-effector motion through Jacobian feedback. A rigorous Lyapunov analysis establishes finite-time convergence and robustness under modeling uncertainties and external disturbances. Extensive experiments-including sinusoidal and square-wave tracking, frequency-sweep tests, and payload variations-demonstrate that the unified controller consistently achieves the lowest tracking errors, superior robustness to excitation frequency and load changes, and smoother actuator effort without increasing energy consumption. The results confirm the suitability of the proposed method for high-precision parallel manipulators operating in dynamic, uncertain, and disturbance-rich environments.