Hierarchical Hybrid Control of Multi-Vehicle Systems

多车辆系统可以看作是既具有离散行为又具有连续行为的混合系统。由于复杂性和功能性的增加，这些系统的开发变得更加困难。此外，这些系统经常在安全关键环境中使用，因此对可靠性要求很高。分布式混合动力系统的设计和实现仍然是一个具有挑战性的任务，本项目研究了协作式无人机分析和综合的关键技术(分层建模的软件工具和基于优化的混合控制)。多车辆系统需要适应手头的任务和来自远程用户的命令。为了便于这类混合系统的设计，提出了以下研究方向：(1)多车辆系统的分层建模。本文的重点是开发一个用于分布式多车辆系统建模的软件工具包。该工具包允许采用模块化和层次化的方法来编程分布式自主操作中的慎重和反应行为。顺序组合、层次组合和并行组合的形式化定义允许自下而上地开发复杂的混合系统。(2)分布式混合控制与优化。这一推力使用最优控制图和模型预测控制(MPC)对一组车辆进行混合控制。确定控制层次结构中应该施加MPC的适当级别是重要的。在极端情况下，MPC可以用于整个系统；然而，当车辆的本地控制器已经很好地建立时，这可能不是有利的。在评估这种权衡时，考虑了算法的复杂性、计算时间和稳定性。在此基础上开发的软件和分析工具具有更广泛的影响，因为它们也可以应用于其他领域，如分布式发电厂、运输系统和编队飞行系统。此外，这些方法有可能适用于商业、军事和民用应用，如矿业、消防和救灾行动、搜索和救援任务，以及国防部任务，如下一代战斗和国土安全。该项目的教育目标是将研究成果整合到本科和研究生课程中，并培养具有这种独特风格的跨学科专业知识的新一代混合和嵌入式系统学生。

Multi-vehicle systems can be viewed as hybrid systems that exhibit both discrete and continuous behaviors. These systems become harder to develop due to increasing complexity and functionality. Furthermore, these systems are frequently used in safety critical environments, and thus, require high assurance in reliability. The design and implementation of distributed hybrid systems remains a challenging task.This project investigates key technologies (software tools for hierarchical modeling, and optimization-based hybrid control) for analysis and synthesis of cooperative unmanned vehicles. Multi-vehicle systems need to adapt to the task in hand and commands from a remote user. To facilitate the design of this class of hybrid systems, the following research thrusts are addressed: (1.) Hierarchical modeling of multi-vehicle systems. This thrust focuses on developing a software toolkit for modeling distributed multi-vehicle systems. This toolkit allows a modular and hierarchical approach to programming deliberative and reactive behaviors in distributed autonomous operation. Formal definitions for sequential composition, hierarchical composition, and parallel composition allow the bottom-up development of complex hybrid systems. (2.)Distributed hybrid control and optimization. This thrust addresses hybrid control of a team of vehicles using optimal control graphs and model predictive control (MPC). Determining the proper level in the control hierarchy at which MPC should be imposed is important. In the extreme, MPC can be used for the entire system; however this may not be advantageous when local controllers for vehicles have already been well established. In evaluating this trade-off, the complexity of the algorithm, computation time, and stability are considered.The software and analysis tools developed under this effort have a broader impact, as they can also be applied to other domains such as distributed power plants, transportation systems, and formation flight systems. Furthermore, these methodologies have the potential to be applicable for commercial, military and civilian applications like mining industry, firefighting and disaster relief operations, search and rescue missions, and DoD missions such as next generation battle warfare, and homeland security. The educational goals of this project are to integrate the research findings into undergraduate and graduate courses and to train a new generation of students on hybrid and embedded systems with cross disciplinary expertise of this unique flavor.