Dielectric elastomer actuators (DEAs) have garnered extensive attention especially in soft robotic applications over the past few decades owing to the advantages of lightweight, large strain, fast response and high energy density. However, because the DEAs suffer from nonlinear elasticity, inherent viscoelastic creep, hysteresis and vibrational dynamics, the modeling, control and self-sensing of DEAs are challenging, thereby hindering the practical applications of DEAs. In order to address these challenges, numerous studies have been conducted. In this review, various physics-based modeling methods and phenomenological modeling methods for predicting the electromechanical response of DEAs are presented and discussed. Different control methods for DEAs are reviewed, which are classified into open-loop feedforward control, feedback control, feedforward-feedback control and adaptive feedforward control. Physics-based self-sensing methods and data-driven self-sensing methods for reconstructing the DEA displacement without the need for additional sensors are discussed. Finally, the existing problems and new opportunities for the further studies are summarized.

翻译：介电弹性体驱动器（DEAs）因其轻质、大应变、快速响应和高能量密度等优势，在过去几十年中，尤其在软体机器人应用中引起了广泛关注。然而，由于DEAs存在非线性弹性、固有粘弹性蠕变、迟滞和振动动力学特性，其建模、控制与自感知具有挑战性，从而阻碍了DEAs的实际应用。为应对这些挑战，已有大量研究开展。本综述介绍并讨论了用于预测DEA机电响应的多种基于物理的建模方法和现象学建模方法，综述了DEA的不同控制方法，包括开环前馈控制、反馈控制、前馈-反馈控制和自适应前馈控制。讨论了无需额外传感器即可重构DEA位移的基于物理的自感知方法和数据驱动的自感知方法。最后，总结了现有问题及未来研究的新机遇。