Modeling, Design and Operation of Robotic Tether-Net Systems for Reliable Capture of Targets

用于可靠捕获目标的机器人系绳网系统的建模、设计和操作

• 批准号: 2128578
• 负责人: Eleonora Botta
• 金额: $ 50.36万
• 依托单位: SUNY at Buffalo
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-09-01 至 2024-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2128578&HistoricalAwards=false
• 关键词: Modeling; Design; Operation; Robotic; Tether

This project will contribute new scientific knowledge related to the design and operation of robotic tether-net systems, thus advancing the national prosperity, welfare and security. The new robotic system consists of a tether-connected net with corner masses that can be launched from a chaser vehicle to capture a moving target object. The launched net can also be actively closed using a variety of embedded mechanisms. A tether-net system provides a potentially safer and more reliable alternative compared to traditional solutions for capturing non-cooperative targets. These advantages could translate to important applications, primarily in the capture and removal of space debris that pose increasing threats to orbital operations, as well as in the recovery of derelict drones. Currently, there is a lack of systematic approaches to study various tether-net configurations and compatible autonomous control paradigms. Moreover, capture has to happen under imperfect sensing of targets of different shapes and tumbling motions, and the effects of these real-world uncertainties on net operation remain poorly understood. This award supports research to address these knowledge gaps by bringing together net dynamics and contact mechanics, engineering optimization and machine learning. Immediate impact of the outcomes of this project will be towards the use-case of space debris removal and thereof continued safe exploitation of commercial orbits. This will benefit satellite operators, as well as U.S. national agencies and the public who rely on earth observation satellites, and will help strengthen U.S. leadership in Space. This multidisciplinary project will also help broaden the participation of underrepresented groups in Science, Technology, Engineering and Math (STEM) and promote exposure of engineering students to the emerging technology of net-based robotics.The inherent structural flexibility of a net presents various under-explored operational paradigms, ranging from passive nets relying on entanglement to complex active solutions with maneuverable corner elements that are equipped with varying degrees of sensing and control capabilities. The overall goal of this research is to develop computational approaches to perform a formal exploration of the resulting distinct operational paradigms and associated design choices that are jointly effective at capturing different types of targets in the presence of uncertainty. To accomplish this research goal, the following key fundamental contributions are envisioned: 1) Develop physics-infused machine learning approaches to model the complex dynamics of the net and its interactions with a target object, at computing-cost/accuracy trade-offs that are desirable for controls and design explorations. 2) Construct, compare and contrast standard centralized control architectures and novel decentralized formation control techniques to regulate the net launch and closure. 3) Explore the interplay of form (net design) and behavior (net control) in affecting capture performance, by adopting a reliability-based optimization process informed by design-adaptive neuro-controllers that will alleviate the computational burden of this concurrent control/design exploration process. These fundamental contributions will be reduced to practice through lab-scale physical experiments with targets such as hanging objects and hovering drones. This, along with the public release of resulting software artifacts that interface with benchmark robot learning environments, will provide a tangible foundation for the robotics community to invent, design and evaluate newer robotic systems based on the tether-net concept. This project is supported by the cross-directorate Foundational Research in Robotics program, jointly managed and funded by the Directorates for Engineering (ENG) and Computer and Information Science and Engineering (CISE).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.

该项目将为机器人绳网系统的设计和操作提供新的科学知识，从而促进国家的繁荣，福利和安全。新的机器人系统由一个带角质量的系绳连接网组成，可以从追逐车辆上发射，以捕获移动的目标物体。启动的网络也可以使用各种嵌入式机制主动关闭。与捕获非合作目标的传统解决方案相比，绳网系统提供了一种潜在的更安全和更可靠的替代方案。这些优势可以转化为重要的应用，主要是在捕获和清除对轨道运行构成越来越大威胁的空间碎片方面，以及在回收无人驾驶飞机方面。目前，缺乏系统的方法来研究各种绳网配置和兼容的自主控制范例。此外，捕获必须发生在对不同形状和翻滚运动的目标的不完美感测下，并且这些现实世界的不确定性对网络操作的影响仍然知之甚少。该奖项支持通过将网络动力学和接触力学、工程优化和机器学习结合起来来解决这些知识差距的研究。 这一项目的成果将直接影响到空间碎片清除的使用情况，并继续安全利用商业轨道。这将有利于卫星运营商以及依赖地球观测卫星的美国国家机构和公众，并将有助于加强美国在太空领域的领导地位。这个多学科项目还将有助于扩大科学、技术、工程和数学（STEM）领域代表性不足的群体的参与，并促进工程专业学生接触基于网络的机器人技术。网络固有的结构灵活性提出了各种未被探索的操作范例，范围从依赖于缠结的无源网到具有配备有不同程度的感测和控制能力的可折叠角元件的复杂有源解决方案。本研究的总体目标是开发计算方法来执行正式的探索所产生的不同的操作模式和相关的设计选择，共同有效地捕捉不同类型的目标，在存在的不确定性。为了实现这一研究目标，设想了以下关键的基本贡献：1）开发注入物理的机器学习方法，以控制和设计探索所需的计算成本/精度权衡，对网络的复杂动态及其与目标对象的相互作用进行建模。2)构建，比较和对比标准的集中式控制架构和新的分散式编队控制技术，以规范网络的发射和关闭。3)探索形式（网络设计）和行为（网络控制）的相互作用，在影响捕获性能，通过采用基于可靠性的优化过程通知设计自适应神经控制器，将减轻这种并发控制/设计探索过程的计算负担。这些基本贡献将通过实验室规模的物理实验，如悬挂物体和盘旋的无人机来实践。这一点，沿着与基准机器人学习环境接口的最终软件工件的公开发布，将为机器人社区发明，设计和评估基于系绳网络概念的更新机器人系统提供切实的基础。该项目由跨部门的机器人基础研究项目支持，该项目由工程部（ENG）和计算机与信息科学与工程部（CISE）共同管理和资助。该奖项反映了NSF的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Learning Robust Policies for Generalized Debris Capture with an Automated Tether-Net System

使用自动化 Tether-Net 系统学习通用碎片捕获的稳健策略

• DOI: 10.2514/6.2022-2379
• 发表时间: 2022
• 期刊: AIAA SCITECH 2022 Forum
• 作者: Zeng, Chen;Hecht, Grant R.;Kumar, Prajit K.;Shah, Raj K.;Botta, Eleonora M.;Chowdhury, Souma
• 通讯作者: Chowdhury, Souma

Auto-differentiable Transfer Mapping Architecture for Physics-infused Learning of Acoustic Field

用于声场物理学习的自微转移映射架构

• DOI: 10.1109/tai.2023.3248561
• 发表时间: 2023
• 期刊: IEEE Transactions on Artificial Intelligence
• 作者: Iqbal, Rayhaan;Behjat, Amir;Adlakha, Revant;Callanan, Jesse;Nouh, Mostafa;Chowdhury, Souma
• 通讯作者: Chowdhury, Souma

Validation of Models for Net Deployment and Capture Simulation with Experimental Data

• DOI: 10.2514/1.a35798
• 发表时间: 2023-08
• 期刊: Journal of Spacecraft and Rockets
• 影响因子: 1.6
• 作者: Achira Boonrath;E. M. Botta

Learning Constrained Corner Node Trajectories of a Tether Net System for Space Debris Capture

学习用于空间碎片捕获的系绳网系统的约束角节点轨迹

• DOI: 10.2514/6.2023-3920
• 发表时间: 2023
• 期刊: AIAA AVIATION 2023 Forum
• 作者: Liu, Feng;Boonrath, Achira;KrisshnaKumar, Prajit;Botta, Eleonora M.;Chowdhury, Souma
• 通讯作者: Chowdhury, Souma