CSR--EHS: Small: Limited-information control of hybrid systems via reachable set propagation

这个项目的主题是基于有限（采样和量化）的混合系统状态信息的控制。该项目旨在首次将两个研究课题结合在一起：混合系统和有限信息控制。该项目将寻求开发一种新的方法来传播混合系统可达集的过近似值，并使用它来定义和验证基于连续状态的采样和量化测量的控制策略。该方法将与最先进的混合和切换系统分析工具相结合，如多重李雅普诺夫函数、平均停留时间切换和输入到状态稳定性。混合系统对于网络物理系统（如电传飞机和电网）的建模和控制行为非常重要。混合动力系统结合了连续动力和离散过渡。这种混合性质使人们能够捕捉到在实践中遇到的在不同操作模式之间切换的许多过程。由于可用传感器的限制以及通信中的延迟和损失，此类系统的控制器通常具有关于系统状态的有限信息。一个可能的场景是控制一个自动机器人，它在搜索模式和追踪模式之间切换，出于安全或实施成本的原因，通过一个非常低功耗的通信链路连接到控制中心。该项目的教育和推广活动包括支持一名研究生，将研究与混合系统和控制的研究生课程、讲习班和短期课程相结合，以及让本科生参与将理论发展转化为软件工具。

The theme of this project is control of hybrid systems based on limited (sampled and quantized) information about their state. The project aims to bring together, for the first time, two research topics: hybrid systems, and control with limited information. The project will seek to develop a novel method for propagating over-approximations of reachable sets for hybrid systems, and use it to define and validate a control strategy based on sampled and quantized measurements of the continuous state. This method will be combined with state-of-the-art analysis tools for hybrid and switched systems, such as multiple Lyapunov functions, average dwell-time switching, and input-to-state stability.Hybrid systems are important for modeling and controlling the behavior of cyber-physical systems, such as fly-by-wire aircraft and power grids. Hybrid systems combine continuous dynamics and discrete transitions. This hybrid nature enables one to capture many processes encountered in practice which switch between different modes of operation. Controllers for such systems typically have limited information about the system state, due to limitations of available sensors and delays and losses in communication. A possible scenario is controlling an autonomous robot which switches between the search mode and the pursuit mode and, for security or implementation cost reasons, is connected to the control center via a very low-power communication link. Educational and outreach activities of the project include support of a graduate student, integration of the research with graduate courses in hybrid systems and control, workshops and short courses, and involvement of undergraduates in translating theoretical developments into software tools.