Finite-time Containment Control for Lagrangian Networks

目的：研究拉格朗日网络的分布式有限时间包容控制问题。尽管遏制控制、网络系统的有限时间控制和拉格朗日网络控制对于网络化多智能体系统来说都是至关重要的，但由于许多固有的挑战，关于每个主题的现有结果很少，更不用说三者的组合了。这项研究在一个统一的框架内解决了这三个主题的组合。智能优点：分布式算法将被设计为保证具有固定领导者、动态领导者和蜂群行为的拉格朗日网络的有限时间遏制控制，并将在异类机器人网络上进行实验验证。研究的创新之处在于范式的变化：i)从多智能体系统中主要研究的无领导或单领导者行为，到更现实但复杂的多领导者多追随者行为，涉及多个不同种类的智能体团队之间的多层协作；ii)从在多智能体系统中主要探索的只能在无限时间内保证渐近或至多指数收敛的分布式算法，到保证在有限时间内收敛以满足许多时间关键任务的需要的分布式算法，从多智能体系统中主要采用的简化/理想化的线性点模型，到更现实、更普遍的非线性拉格朗日模型，这些模型代表了包括自动驾驶车辆和机器人操作手在内的一大类机械系统。广泛的影响：该研究对许多民用、国土安全和军事应用以及与生物、社会学和物理等网络系统相关的其他领域产生了影响。将开发一门关于自主系统的顶峰课程，重点是动手实验室和全校范围的机器人比赛。针对高中生的外展活动将通过参与犹他州工程州立大学的全州工程项目来进行。将通过出版物、辅导讲习班、暑期学校和专门网站广泛传播研究成果。

Objective:The objective of this research is to address the problem of distributed finite-time containment control for Lagrangian networks. Despite the fact that each topic of containment control, finite-time control of networked systems, and control of Lagrangian networks is crucial for networked multi-agent systems, there are few existing results on each topic not to mention a combination of the three due to numerous inherent challenges. This research addresses a combination of the three topics in a unified framework.Intellectual Merit:Distributed algorithms will be designed to guarantee finite-time containment control for Lagrangian networks with stationary leaders, dynamic leaders, and swarming behavior and will be experimentally validated on heterogeneous robotic networks. The novelty of the research lies in the paradigm changes i) from leaderless or single leader behaviors dominantly studied in multi-agent systems to more realistic but complicated multi-leader multi-follower behaviors involving multi-layer collaboration between multiple heterogeneous teams of agents, ii) from distributed algorithms that can only guarantee asymptotic or at most exponential convergence in infinite time dominantly explored in multi-agent systems to distributed algorithms that are guaranteed to converge in finite time to meet the needs of many time-critical missions, and iii) from simplified/idealized linear point models dominantly adopted in multi-agent systems to more realistic and prevalent nonlinear Lagrangian models that represent a large class of mechanical systems including autonomous vehicles and robotic manipulators.Broader Impacts: The research has impact on numerous civilian, homeland security, and military applications and on other fields related to networked systems including biology, sociology, and physics. A capstone course on autonomous systems will be developed with emphasis on hands-on labs and campus-wide robotics competitions. Outreach activities to high school students will be pursued through involvement in a state-wide engineering program in Utah, Engineering State. Broad dissemination of the research results will be accomplished through publications, tutorial workshops, a summer school, and a dedicated website.