Robust Control of Shape Memory Alloys for Assistive Robotics Applications

Shape Memory Alloys (SMAs) are a popular class of actuators widely used in complex soft robotics applications due to their shape memory effect, high recoverable strain, and stress. However, most existing actuation models frequently fail to accurately capture hysteresis and dynamic loading behavior while remaining computationally efficient. Moreover, current control strategies often lack adaptability, robustness, and the ability to generalize to varying system dynamics. This paper presents a robust adaptive closed-loop controller for electro-thermally actuated Ni–Ti SMAs, developed based on a Finite State Machine framework to address these challenges. The proposed controller is designed to compensate for disturbances and uncertainties in the SMA behavior. Experimental validation and statistical analysis have demonstrated the effectiveness of the L₁ L 1 adaptive controller across various SMA configurations, enabling precise strain and stress target tracking. Finally, the controller is deployed to a case study involving a Ni–Ti SMA-powered assistive robotic device, where it successfully manages position tracking with enhanced performance.