Analysis and Verification on Backstepping Control of Antagonistic Variable Stiffness Actuators

Abstract

Variable stiffness actuators (VSAs) are an important class of compliant actuators that can benefit the robustness and task adaptability of robotic systems. However, controlling VSAs is still challenging as VSAs are difficult to model exactly and have highly nonlinear characteristics. This study applies a model-based robust command-filtered backstepping control (CFBC) approach to the position tracking of agonistic-antagonistic (AA) VSAs and qualitatively analyzes how model uncertainties affect performance. Simulations and experiments based on an AA-VSA named qbmove Advanced are carried out for validation. It is revealed that model accuracy is still critical for robust CFBC of AA-VSAs, where the inaccuracy of inertia, friction, and elastic parameters deteriorates the control performance. Specifically, the unavailability of elastic torque measurement makes model-based control impractical for AA-VSAs under model uncertainties.