本文研究了线性时不变系统中最小 actuators 的可达性问题,提出零一对角输入矩阵 B 的构造方法,证明前两个问题为 NP 难,并提供多项式时间算法及性能验证。
摘要:
本文针对线性时不变系统中的状态空间可达性问题,使用最少的执行器设计零一对角输入矩阵 B,使得指定状态向量从给定初始状态可达。此外,设计 B 使系统能被引导至给定子空间或接近期望状态。本文扩展了 Olshevsky 和 Pequito 的最新结果,证明前两个问题为 NP 难,并提供多项式时间算法及最坏情况近似保证,最后在大规模随机网络上展示了性能。
英文摘要:
In this paper, we address a collection of state space reachability problems, for linear time-invariant systems, using a minimal number of actuators. In particular, we design a zero-one diagonal input matrix B, with a minimal number of non-zero entries, so that a specified state vector is reachable from a given initial state. Moreover, we design a B so that a system can be steered either into a given subspace, or sufficiently close to a desired state. This work extends the recent results of Olshevsky and Pequito, where a zero-one diagonal or column matrix B is constructed so that the involved system is controllable. Specifically, we prove that the first two of our aforementioned problems are NP-hard; these results hold for a zero-one column matrix B as well. Then, we provide efficient polynomial time algorithms for their general solution, along with their worst case approximation guarantees. Finally, we illustrate their performance over large random networks.
pyspect: An Extensible Toolbox for Automatic Construction of Temporal Logic Trees via Reachability Analysis
Division of Decision and Control Systems, EECS, KTH Royal Institute of Technology(皇家理工学院电气计算机工程与计算机科学系决策与控制系统分部);Integrated Transport Research Lab(综合交通研究实验室);Digital Futures(数字未来中心);Division of Robotics, Perception and Learning, EECS, KTH Royal Institute of Technology(皇家理工学院电气计算机工程与计算机科学系机器人、感知与学习分部)
Institute of Applied Mathematics and Mechanics, NAS of Ukraine(乌克兰国家科学院应用数学与力学研究所);Department of Computer Science, University of Liverpool, UK(利物浦大学计算机科学系)
On the complexity of bounded time and precision reachability for piecewise affine systems
ENS Cachan/Bretagne et Université Rennes 1(巴黎高等师范学校卡尚分校/布列塔尼与雷恩第一大学);École Polytechnique(巴黎综合理工学院);Egypt Japan University of Science and Technology(埃及日本科学技术大学);Faculty of Engineering, Alexandria University(亚历山大大学工程学院)
On the Decidability of Reachability in Linear Time-Invariant Systems
CNRS(法国国家科学研究中心);LaBRI(计算机科学研究与教学联盟实验室);Alan Turing Institute of Data Science(图灵数据科学研究所);Max Planck Institute for Software Systems(马克斯·普朗克软件系统研究所);IRIF(信息与随机相互作用研究所);Université Paris Diderot(巴黎第七大学);Department of Computer Science, University of Oxford(牛津大学计算机科学系)
DeepReach: A Deep Learning Approach to High-Dimensional Reachability
University of California, Berkeley(加州大学伯克利分校)
Hamilton Jacobi(HJ)可达性分析是保证动态控制系统性能和安全性的一种重要形式化验证方法。它的优点包括与一般非线性系统动力学的兼容性,有界扰动的形式化处理,以及处理状态和输入约束的能力。然而,它涉及到求解偏微分方程,其计算和内存的复杂性相对于状态变量的数量呈指数级增长,限制了其直接用于小规模系统。我们提出了DeepReach,一种利用正弦网络的新发展来开发高维可达性问题的神经PDE求解器的方法。DeepReach的计算需求不直接与状态维度成比例,而是与底层可达管道的复杂性成比例。DeepReach实现了与最先进的可达性方法相当的结果,不需要对PDE解决方案进行任何显式监督,可以轻松处理外部干扰,对抗性输入和系统约束,并且还为系统提供了安全控制器。我们在一个9D多车碰撞问题和一个10D狭窄通道问题上演示了DeepReach,这是由自动驾驶应用引起的。