CAREER: A Compositional Approach to Modular Cyber-Physical Control System Design

职业：模块化网络物理控制系统设计的组合方法

• 批准号: 1553873
• 负责人: Necmiye Ozay
• 金额: $ 50万
• 依托单位: Regents of the University of Michigan - Ann Arbor
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2016
• 资助国家: 美国
• 起止时间: 2016-02-15 至 2022-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1553873&HistoricalAwards=false
• 关键词: CAREER; Compositional; Approach; Modular; Cyber

Complex, networked, distributed cyber-physical systems (CPSs) are emerging in many safety-critical application domains such as aerospace and automotive. Design of such systems heavily relies on insights and experiences of engineers as principled design methodologies that can cope with the complexity of these systems are lacking. As a result, extensive testing and fine-tuning is required to ensure that the final product satisfies the design objectives. As a principled alternative, this project proposes to use modularity for managing complexity during both the design- and the life-cycles of cyber-physical systems. The objective is to develop the scientific foundation and associated algorithmic tools for the design of modular cyber-physical control systems. If successful, in the long-run this research will lead to a "plug and play" integration framework for CPSs supported by automated design tools, where one can replace a subsystem with another one or perform upgrades to subsystems while maintaining operational correctness guarantees. Results from this research will be relevant to many application domains, including next generation air vehicles, automotive systems and robotics. Its potential transformative impact will be on the way CPSs in these domains are designed and operated. Translation to the economy will proceed by actively seeking and engaging industrial partners. This research effort will be complemented by an education plan where interdisciplinary research and thinking in the area of CPS will be fostered among undergraduate and graduate students to prepare the next generation of CPS researchers and practitioners.To be specific, the project will develop theoretical foundations and associated algorithmic tools for distributed synthesis of provably correct control protocols that give rise to compositional design principles for cyber-physical control systems. In particular, algorithms for decompositions of system requirements at the discrete/logic level and of the system states at the continuous/system level will be developed. The main idea is a novel separation between external and internal factors affecting each subsystem that allows internal interactions required for the successful operation of a subsystem to be computed explicitly. These internal interactions, namely interface rules, are captured in terms of assumption and guarantee pairs that are used for solving local synthesis problems to obtain local controllers in a distributed manner, while maintaining global correctness guarantees when these controllers are deployed simultaneously. The modularity-performance trade-off space will be explored by introducing proper partial orders on these interface rules and by tuning the complexity of the interface rules according to these order relations. Tools from control theory (decentralized and robust control, model reduction, discrete event systems) and formal methods (temporal logics, compositional verification, distributed reactive synthesis) will be brought to bear to address these problems.

复杂的、网络化的、分布式的信息物理系统（CPS）正在许多安全关键应用领域中出现，如航空航天和汽车。这种系统的设计严重依赖于工程师的见解和经验，因为缺乏能够科普这些系统复杂性的原则性设计方法。因此，需要进行广泛的测试和微调，以确保最终产品满足设计目标。作为一个原则性的替代方案，该项目建议在网络物理系统的设计和生命周期中使用模块化来管理复杂性。目标是为模块化网络物理控制系统的设计开发科学基础和相关的算法工具。如果成功的话，从长远来看，这项研究将导致一个“即插即用”的自动化设计工具支持的CPS集成框架，其中一个可以取代一个子系统与另一个或执行子系统的升级，同时保持操作的正确性保证。这项研究的结果将与许多应用领域相关，包括下一代飞行器，汽车系统和机器人技术。其潜在的变革性影响将是这些领域的CPS的设计和运作方式。将通过积极寻求和参与工业合作伙伴来实现经济转化。这项研究工作将得到一项教育计划的补充，该计划将在本科生和研究生中培养CPS领域的跨学科研究和思维，为下一代CPS研究人员和从业人员做好准备。该项目将开发理论基础和相关的算法工具，用于分布式合成可证明正确的控制协议，从而产生网络的组合设计原则，物理控制系统。特别是，将开发用于在离散/逻辑级和在连续/系统级的系统状态的系统要求的分解的算法。其主要思想是一种新的外部和内部因素之间的分离影响每个子系统，允许内部的相互作用所需的子系统的成功运行进行明确计算。这些内部的相互作用，即接口规则，被捕获的假设和保证对，用于解决本地合成问题，以获得本地控制器在分布式的方式，同时保持全局正确性的保证，当这些控制器同时部署。通过在这些接口规则上引入适当的偏序，并根据这些序关系调整接口规则的复杂性，将探索模块化性能权衡空间。控制理论（分散和鲁棒控制，模型简化，离散事件系统）和形式化方法（时态逻辑，组合验证，分布式反应合成）的工具将被用来解决这些问题。

项目成果

Scalable Computation of Controlled Invariant Sets for Discrete-Time Linear Systems with Input Delays

具有输入延迟的离散时间线性系统的受控不变集的可扩展计算

• DOI: 10.23919/acc45564.2020.9147731
• 发表时间: 2020
• 期刊: 2020 American Control Conference (ACC
• 作者: Liu, Zexiang;Yang, Liren;Ozay, Necmiye
• 通讯作者: Ozay, Necmiye

Finite horizon constrained control and bounded-error estimation in the presence of missing data

存在缺失数据时的有限视野约束控制和有界误差估计

• DOI: 10.1016/j.nahs.2020.100854
• 发表时间: 2020
• 期刊: Nonlinear Analysis: Hybrid Systems
• 作者: Rutledge, Kwesi;Yong, Sze Zheng;Ozay, Necmiye
• 通讯作者: Ozay, Necmiye

Tight decomposition functions for mixed monotonicity

混合单调性的紧分解函数

• 发表时间: 2019
• 期刊: 58th IEEE Conference on Decision and Control (CDC
• 作者: Yang, Liren;Ozay, Necmiye
• 通讯作者: Ozay, Necmiye

Model Error Propagation via Learned Contraction Metrics for Safe Feedback Motion Planning of Unknown Systems

• DOI: 10.1109/cdc45484.2021.9683354
• 发表时间: 2021-04
• 期刊: 2021 60th IEEE Conference on Decision and Control (CDC)
• 作者: Glen Chou;N. Ozay;D. Berenson

Uncertainty-Aware Constraint Learning for Adaptive Safe Motion Planning from Demonstrations

通过演示进行自适应安全运动规划的不确定性约束学习

• 发表时间: 2020
• 期刊: Conference on Robot Learning
• 作者: Chou, Glen;Ozay, Necmiye;Berenson, Dmitry
• 通讯作者: Berenson, Dmitry