Modeling and Control of Soft Robots Actuated by Dielectric Elastomer

在过去的十年中，软机器人因其潜在的应用受到了极大的关注。软体机器人主要是指采用柔软或高度可变形的材料，具有柔软或柔性结构和类肌肉驱动的机器人系统。与传统机器人不同的是，它们可以在驱动下表现出高度变形，并沿着身体长度不断改变形状，类似于章鱼的手臂、蚯蚓和象鼻。软驱动技术的最新进展为软机器人的发展提供了有利条件。在软致动材料中，介电弹性体（DEs），也被称为介电活性聚合物，代表了一类新兴/创新的软致动材料，当被电场激发时表现出相对较大的变形。DEs由弹性聚合物材料薄膜组成，薄膜两侧覆盖有柔性电极。当对电极施加电压时，所产生的电场产生压应力，从而使薄膜产生可控的变形。在某些情况下，这种变形可能比其他软材料的变形大一到两个数量级。这种特性，再加上其重量轻、能量密度高、响应速度快、成本低等特点，使其对软机器人的发展具有极大的吸引力。另一方面，完全或部分由可改变形状的材料制成的机器人比刚性机器人更难控制。DEs是粘弹性材料，这意味着它们的刚度随应变速率和频率而变化。粘弹性使材料响应的建模和控制复杂化，而材料响应通常以蠕变、迟滞和应力松弛现象为特征。由于这些固有的非线性和时变特性，如何精确控制微分方程以确定获得软机器人变形所需的电压量成为一个不容小视的问题。特别是迟滞非线性会显著降低作动器的性能。然而，由于绝大多数现有的控制方法只适用于光滑可微的系统，因此补偿滞后是一个非常困难的问题。应力松弛也是DEs的一个重要缺点，由于弹性体厚度的变化，随着时间的推移，应力松弛会显著降低致动器的性能。这种时变的性能特征很难建模，并提出了具有挑战性的控制问题。本研究的目的是发展DE驱动软机器人的控制理论、工程设计方法和技术，使其适应周围环境，在非结构化环境中导航，并在存在模型不确定性的情况下使用全臂操作来抓取物体。

Over the last decade, soft robots have received tremendous interests for their potential applications. Soft robots mostly refer to robotic systems with soft or compliant structures and muscle-like actuation using soft or highly deformable materials. Unlike conventional robots, they can show high deformation under actuation and change their shapes continuously along their body length, similar to octopus arms, earthworms, and elephant trunks. Recent advances in soft actuation technologies have been favorable to development of soft robotics. Among soft actuation materials, dielectric elastomers (DEs), also known as dielectric electroactive polymers, represent an emerging/innovative class of soft actuation materials which exhibit relatively large deformations when solicited by an electric field. DEs consist of a film of elastic polymeric material covered on both sides by compliant electrodes. When a voltage is applied to the electrodes, the resulting electric field generates a compressive stress that produces a controllable deformation of the film. This deformation can be in some cases one or two orders of magnitude larger than the one obtained by other soft materials. This peculiarity, together with the lightweight, the high energy density, the fast response, and the low costs, makes DEs highly attractive for development of soft robots.On the other hand, robots made wholly or in part from materials that change the shape are more difficult to control than rigid robots. DEs are viscoelastic materials meaning that their stiffness changes with strain rate and frequency. Viscoelasticity complicates the modeling and control of the material response which is often characterized by creep, hysteresis, and stress relaxation phenomena. Owing to these intrinsic nonlinear and time variant characteristics, the precise control of DEs to determine the amount of voltage needed to obtain the deformation for soft robots becomes a non-trivial problem. In particular, hysteresis nonlinearity can significantly degrade actuator performance. However, compensating the hysteresis is a very difficult problem because the vast majority of existing control methods are only applicable to systems that are smooth and differentiable. Stress relaxation is also a significant shortcoming of DEs and can significantly degrade actuator performance over time owing to changes in elastomer thickness. Such time-variant performance characteristics are difficult to model and present challenging control issues. The objective of this research is the development of control theories, engineering design methods and technology for DE actuated soft robots to conform to their surroundings, navigate through unstructured environments, and grasp objects using whole arm manipulation in the presence of model uncertainties.