Practical Stability of Networked Control Systems under Uncertainty

University Of California-San Diego, La Jolla CA

Investigators

Abstract

This project addresses the gap between theory and practice in the control of networked systems in real-world settings. In many practical systems, like autonomous vehicles or assisted driving in smart cities, physical inputs from sensors are converted into digital form and communicated to a controller that determines the appropriate action to take, at any given time of the system evolution, based on the current estimate of the state. One of the main issues in these systems is how to ensure the correct execution of the control task, despite several uncertainties that are present in the control loop. For example, the system can be subject to random disturbances, and the communication channel can be affected by random errors or delays. Despite these stochastic events, it is required that the system maintains a high level of reliability. This is often a stringent requirement since it is directly related to human safety. To successfully merge theory with practice, the project aims at providing a practically relevant analysis of the probabilistic guarantees that the system can provide; and at designing systems based on probabilistic safety for given operating conditions. The study has implications beyond the field of control and communications, in terms of the introduction of new mathematical methods and design tools. In laying a theoretical foundation, we expect to draw novel, synergistic connections between information theory, control theory, and real-time systems. The framework advocated in this proposal will benefit several application domains in need of verifiable designs with provably correct guarantees on performance, such as autonomous transportation, smart power systems, and robotics. The proposed research will also impact the training of a new generation of students through undergraduate student involvement, graduate mentoring and curriculum development, and outreach activities. The goals of the project are as follows: (i) develop a theory suitable to a practically relevant description of system guarantees; (ii) develop system design approaches that embrace probabilistic safety for the given operating conditions; (iii) provide development strategies aimed at satisfying the operating conditions that guarantee probabilistic safety. The theory will include deterministic as well as stochastic analysis, time triggering and event-triggering designs. It will also account for the information implicit in the timing of the communication events and for nonlinearities arising in the plant modeling in the context of event-triggering control. The results of the project will be tested via software simulations, hardware-in-the-loop validation, and experiments on a team of aerial and ground vehicles at the UCSD AeroDrome. Core application areas include autonomy of transportation and smart cities, highly dynamical systems, convergence of computation, communication, and control, environment and infrastructure monitoring, and smart power systems. The intellectual merit of the proposal is to provide sound reliability guarantees on the operation of networked control systems where communication and control aspects system are treated in an integrated fashion. The proposed theory will consider notions of stability that are attractive from a practical perspective. Depending on the type of communication channel and system disturbances, our results will range from deterministic, to strong probabilistic, or weak probabilistic guarantees on the ability to successfully achieve the control objective. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.