Comparative Analysis of Model-Based and Model-Free Control Strategies for Viscoelastic Dielectric Elastomer Actuator

Soft actuators, fabricated with soft materials, have garnered significant attention for their intrinsic compliance and adaptability. Among them, dielectric elastomer actuators (DEAs) have been widely applied due to their high power density, lightweight design, and quiet operation. Despite these advantages, the viscoelastic behaviors of DEAs, including hysteresis and creep, pose substantial challenges to accurately modeling and controlling their nonlinear mechanical responses. This study presents a comparative analysis of two control methodologies: a model-based controller and a model-free controller. The model-based approach employs sparse identification methods to derive the dynamic equation, followed by the design of a nonlinear model predictive controller (NMPC). In contrast, the model-free approach utilizes a proportional-integral-derivative (PID) controller for motion regulation. The performance of both controllers is systematically evaluated based on their ability to track sinusoidal trajectories across various frequencies. The NMPC achieves superior tracking accuracy, while the PID strikes a balance between tracking performance and simplicity in controller design. The findings provide critical insights into the selection of appropriate control strategies for DEA-based soft actuators and robotic systems, advancing their application in soft robotics and beyond.