Simulation study on prescribed-time stabilization of 5-DOF exoskeletons using a regularized super-twisting approach

Abstract This paper proposes an enhanced Prescribed-Time Super-Twisting Controller (PT-STC) for 5-DOF upper-limb rehabilitation exoskeletons. A regularized scaling transformation is introduced to ensure tracking error convergence to a tunable $$\varepsilon$$ -neighborhood of the origin within a user-defined time window, independent of initial conditions and disturbance magnitudes. The terminal error bound is characterized as $$\Vert e(T_{user})\Vert \le \mathcal {O}(\varepsilon )$$ , providing a systematic trade-off between convergence precision and control effort through the selection of $$\varepsilon$$ . Unlike conventional prescribed-time approaches, the proposed “Soft-Landing” mechanism eliminates gain explosion singularities, thereby preventing actuator saturation and maintaining control signals within safe operational limits in simulation. The integration of the Super-Twisting Algorithm within the scaled coordinate domain yields chattering-free torque profiles essential for safe human-robot interaction. The theoretical developments are validated through high-fidelity simulations under nominal stabilization and dynamic tracking with impact disturbances. Results demonstrate a settling time of approximately 1.99s, with significant reductions in both settling time and total variation relative to conventional sliding mode control. These findings suggest that the PT-STC offers a promising balance between temporal precision and smooth actuation, warranting further experimental investigation. We emphasize that the current results are simulation-based; experimental validation is required before clinical deployment.