CAREER: Control with Limited Information

本研究的重点是理解控制与信息之间的相互作用，并将研究结果应用于通信网络、动态视觉和自动驾驶汽车的控制问题。这些应用程序都有一个共同的约束，即与信息相关的成本，由于资源有限，限制了可用信息的控制。这些应用都是混合系统的例子。了解有限的信息如何影响控制系统可实现的性能是本研究的主要目标。如果我们能够学会如何在不失去闭环系统稳定性和性能的情况下系统地量化信息，我们可以设计出混合系统分析、控制和设计的新方法，并直接影响许多应用领域。这些新问题需要多学科研究，依赖于不确定系统控制的丰富经验，采用优化方法的新进展，并将其与信息论概念相结合。基于这一方法的初步研究已经为线性系统的稳定性和性能保证提供了新的系统理论结果。这些结果非常有希望，表明这项研究可以取得巨大的进步。激励应用和新的理论问题的混合提供了一个独特的教育机会，学生将参与该计划。该研究计划的多学科性质将为将开发的两门课程提供材料和动力。第一门课程面向本科生，侧重于系统设计，特别是控制、通信和微处理器系统的集成。第二部分侧重于系统设计、分析和控制的优化方法，并将优化理论和计算方法作为跨电气工程学科的统一工具。

The proposed research focuses on understanding the interaction between control and information, and applying the findings to the problems of control over communication networks, dynamic vision, and autonomous vehicles.These applications have in common the constraint that thereis a cost associated with information, which, due to finite resources, limits the available information for control. These applications are examples of hybrid systems.Understanding how the limited information affects the performance achievable by control systems is a primary goal of this research. If we can learn how to systematically quantize the information without loosing stability or performance of the closed loop system, we can devise new methods of analysis, control, and design of hybrid systems, and directly impact many application fields.These new problems require a multidisciplinary approach that relies on the rich experience in control of uncertain systems, adopts the new advances in optimization methods, and merges them with information theory concepts.Preliminary research based on this approach has led to new system theory results for stability and performance guarantee with quantized state information for linear systems. These results are very promising and indicate that vast advances can be made by this study.The mix of motivating applications and new theoretical problems offers a unique educational opportunity to the students that will be involved in the program.The multidisciplinary nature of the research program will provide material and motivation for two courses that will be developed. The first one, intended for undergraduates, concentrates on system design, particularly on the integration of control, communication, and microprocessors systems.The second one focuses on optimization methods for system design, analysis, and control, and presents optimization theory and computational methods as unifying tools across Electrical Engineering disciplines.