Towards Sensorless Soft Robotics: Self-Sensing Stiffness Control of Dielectric Elastomer Actuators

In this article, we present a sensorless stiffness control (SC) architecture for a soft dielectric elastomer (DE) membrane actuator. The method relies on a self-sensing algorithm that exploits measurements of electrical quantities (i.e., membrane voltage and current) to perform a real-time estimation of DE displacement and force. By combining self-sensing feedback with a SC algorithm, active shaping of the membrane force-displacement response is achieved without introducing additional electro-mechanical sensors in the system, thus, making it possible to design compact, lightweight, and low-cost DE robotic systems. A description of the novel self-sensing scheme is initially performed. To cope with the strong system nonlinearity, a robust design method to synthesize a SC law is subsequently proposed. An extensive experimental campaign is, then, carried out, with the goal of evaluating the performance of both sensor-based and sensorless SC. Quantitative accuracy of both control architectures is finally assessed and compared.