March 14, 2018
332 McBryde Hall
Jorge Poveda, University of California, Santa Barbara
Abstract: The deployment of real-time control and optimization strategies in cyber-physical systems could significantly improve our quality of life while creating jobs and economic opportunity. However, cyber-physical systems, such as autonomous vehicles, robotic systems, transportation networks, healthcare systems, and smart grids, are continuously exposed to potential threats and attacks generated by adversarial entities. These circumstances have made safety and security a top priority in the development of feedback mechanisms for cyber-physical systems. However, given the highly complex interactions between physical and digital processes that emerge in this type of systems, the development of robust, reliable, and safe feedback mechanisms requires novel analytical and algorithmic tools that combine ideas from nonlinear robust control theory, hybrid dynamical systems, optimization, and game theory.
In this talk, I will discuss how the framework of robust set-valued hybrid dynamical systems can help us achieve such a goal by providing a convenient analytical framework to design data-driven feedback mechanisms that retain their stability properties under a general class of undetectable adversarial perturbations, and by providing a powerful methodology to model families of adversarial attacks in dynamical systems. Specically, in the rst part of the talk I will present a novel analytical framework for the design of a family of adaptive hybrid control algorithms that robustly optimize the performance of a hybrid system under the eect of undetectable adversarial attacks acting on the states and dynamics of the closed-loop system. In the second part of the talk, I will show how the framework of set-valued stochastic hybrid systems can be used to construct a class of stochastic feedback mechanisms designed to control a network of multiple cyber-physical systems interacting in a non-cooperative game, and operating in an environment where external attacks persistently deactivated the feedback mechanisms of some of the
nodes of the network. By using the framework of set-valued dynamic inclusions and novel stability results for stochastic systems under persistent perturbations, I will show that, provided a causality condition is satised by the attacks, a family of mitigating conditions that preserve the stability properties of the closed-loop system can be established. I will conclude with a brief outlook on future research directions.
Bio: Jorge I. Poveda is a Ph.D. Candidate at the Center for Control, Dynamical Systems, and Computation (CCDC) at the University of California, Santa Barbara. He received the B.S. degrees in Electronics Engineering and Mechanical Engineering in 2012, and the M.S. degree (Magna Cum Laude) in Electrical Engineering in 2013, all from University of Los Andes, Bogota, Colombia, and the M.S. degree in Electrical and Computer Engineering from the University of California, Santa Barbara, USA, in 2015. He was a Research Intern with the Mitsubishi Electric Research Laboratories in Cambridge, MA, during the summers of 2016 and 2017. He received the 2013 CCDC Outstanding Scholar Fellowship at UCSB, and was a nalist for the Best Student Paper Award at the 56th IEEE Conference on Decision and Control in
2017. His main research interests lie at the intersection of robust hybrid
dynamical systems, online robust optimization, and game theory, with applications to cyber-physical and socio-technical systems.