Towards Accurate and Efficient Dynamics Modeling and Control for Soft Robots in Unstructured Environments

近年来，随着对先进机器人系统在非结构化环境下高效、稳健、智能和安全运行的需求不断增加，软机器人领域发展迅速。与传统的刚性机器人系统相比，柔性机器人具有高度的灵活性，因为它们可以变形其柔软的身体以适应集群环境而不会损坏或堵塞。此外，软机器人还具有低成本、低功耗、轻量化和对人类友好的特点，可以促进许多原本无法实现的应用，例如在农业收获中拾取易碎物体（例如西红柿），在医疗应用中的非侵入性内窥镜手术，以及在深海勘探中的软机器人抓取器等。然而，软机器人在实际应用中的实现在许多技术方面提出了挑战，包括材料、传感、建模和控制。软机器人的研究还处于初级阶段，缺乏一个统一的框架来设计、建模和控制高灵巧的软机器人。该项目支持基础研究，以提供填补这一重要技术空白所需的知识，从而促进软机器人在空间/水下探测、航空航天、医疗保健、生物医学和农业等行业的广泛应用。因此，本研究成果不仅将促进机器人科学与工程的进步，而且将有利于美国的经济、社会和国防。这项研究涉及多个学科，包括机器人、控制理论、机器学习、计算力学和材料科学。跨学科的方法将有助于促进少数民族参与科学研究和促进工程教育。本项目的目标是研究一种在非结构化环境下柔性机器人操作和运动的高效、准确的动力学建模和控制框架。具体来说，研究团队将开发具有分布式动力、驱动和触觉传感的模块化即插即用软机械臂。建立并实现了一个三维、降阶、几何精确、具有肌腱收缩内部驱动力的柔性机械臂有限元模型。为软体机器人与环境相互作用的动力学特性开发一套高效、准确的仿真程序。动态控制器，混合基于模型和数据驱动的控制策略，也将设计和测试。通过一系列的模拟和物理实验，该团队将确定基本的设计原则，并开发控制器，以实现自主软机器人，如象鼻子一样灵巧的软操纵器和章鱼一样的行走/游泳软机器人。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The field of soft robotics is growing rapidly in recent years, with the rising demand for advanced robotic systems to operate efficiently, robustly, intelligently, and safely under unstructured environments. Compared to traditional rigid robotic systems, soft robots are highly dexterous in the sense that they can deform their soft body to adapt to clustered environments without damage or jamming. In addition, soft robots can be made to be low-cost, low-power, lightweight, and human friendly, promoting many applications that are otherwise not available, such as picking up fragile objects (e.g., tomatoes) in agriculture harvesting, non-invasive endoscopic surgery in medical applications, and soft robotic grippers in deep-sea exploration, etc. However, implementation of soft robots in real-world applications presents challenges in many technological aspects, including material, sensing, modeling, and control. The research of soft robotics is still in its primitive stage as a unified framework for the design, modeling, and control of highly dexterous soft robots is still lacking. This project supports fundamental research to provide the knowledge needed to fill this important technology void, so as to promote a wide range of soft robot applications, e.g., in-space/underwater exploration, aerospace, healthcare, biomedical, and agricultural industries. Therefore, results from this research will not only promote the progress of robotics science and engineering, but also benefit the U.S. economy, society, and national defense. This research involves multiple disciplines including robotics, control theory, machine learning, computational mechanics, and material science. The interdisciplinary approach will help boost minority involvement in scientific research and promote engineering education.The goal of this project is to investigate an efficient and accurate dynamics modeling and control framework for soft robotic manipulation and locomotion under unstructured environments. Specifically, the research team will develop modular plug-and-play soft robotic arms with distributed power, actuation, and tactile sensing. A three-dimensional, reduced-order, geometrically exact, finite element model of soft robotic arms with internal actuation forces from tendon contraction will be formulated and implemented. An efficient and accurate simulation program for the dynamics of soft robots interacting with the environments will be developed. Dynamic controllers, hybridizing model-based and data-driven control strategies, will also be devised and tested. Through a series of simulations and physical experiments, the team will identify fundamental design principles and develop controllers for enabling autonomous soft robots such as elephant trunk-like dexterous soft manipulators and octopus-like walking/swimming soft robots.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.