Data Driven Learning and Control seminar series is organized by the Information and Decision Science Lab at Cornell University and aims to explore the latest advancements and interdisciplinary approaches to data-driven learning and control systems.
Watch on YouTube Live
Bridging Models and Data: Enhancing Cyber-Physical System Security and Control
Mathematical models have long played a crucial role in the design and operation of cyber-physical systems (CPSs), enabling practitioners to verify essential system properties, identify vulnerabilities, and synthesize performance-specific controllers offline. These models also provide a quantitative foundation for assessing core system attributes, such as controllability and observability-properties that can be targeted or compromised by malicious actors. In the first part of this talk, we will introduce novel concepts for quantifying a system’s distance to uncontrollability and unobservability under attack, framed as functions of the CPS model. Our proposed metrics of controllability and observability can guide modifications to system architecture, such as actuator and sensor selection, to enhance resilience against adversarial threats. While accurate models are invaluable to system operators, their effectiveness hinges on precision. In practice, modeling errors are often non-trivial and cannot be ignored, particularly for performance-critical tasks. A natural solution to this challenge is the integration of model-free learning methods, which leverage real-time trajectory data to complement traditional model-based approaches. In the latter half of this talk, we will focus on a concrete example: the optimal regulation problem. We will present a framework that combines model-based and model-free controllers, intelligently transitioning from a purely model-based controller to a composite controller that incorporates an auxiliary component derived via model-free reinforcement learning. Through this series of works, we aim to both extend and exploit the capabilities of mathematical models for CPSs, while judiciously augmenting model-based architectures with model-free techniques when necessary.
Bio: Kyriakos G. Vamvoudakis was born in Athens, Greece. He earned his Diploma in electronic and computer engineering (equivalent to a Master of Science) from the Technical University of Crete, Greece, in 2006, graduating with highest honors. After relocating to the United States, he pursued further studies at the University of Texas at Arlington under the guidance of Frank L. Lewis, obtaining his M.S. and Ph.D. in electrical engineering in 2008 and 2011, respectively. From May 2011 to January 2012, he served as an adjunct professor and faculty research associate at the University of Texas at Arlington and the Automation and Robotics Research Institute. Between 2012 and 2016, he was a project research scientist at the Center for Control, Dynamical Systems, and Computation at the University of California, Santa Barbara. He then joined the Kevin T. Crofton Department of Aerospace and Ocean Engineering at Virginia Tech as an assistant professor, a position he held until 2018.
He currently serves as the Dutton-Ducoffe Endowed Professor at The Daniel Guggenheim School of Aerospace Engineering at Georgia Tech. He holds a secondary appointment in the School of Electrical and Computer Engineering. His expertise is in reinforcement learning, control theory, game theory, cyber-physical security, bounded rationality, and safe/assured autonomy.
He has received numerous prestigious honors, including the 2019 ARO YIP Award, the 2018 NSF CAREER Award, the 2018 DoD Minerva Research Initiative Award, and the 2021 Georgia Tech Chapter Sigma Xi Young Faculty Award. His research has also earned multiple best paper nominations and international recognitions, such as the 2016 International Neural Network Society (INNS) Young Investigator Award, a 2024 NASA Group Achievement Award, the Best Paper Award for Autonomous/Unmanned Vehicles at the 27th Army Science Conference (2010), the Best Presentation Award at the World Congress of Computational Intelligence (2010), and the Best Researcher Award from the Automation and Robotics Research Institute (2011). Dr. Vamvoudakis has actively contributed to the research community through service on numerous international program committees and by organizing special sessions, workshops, and tutorials at major international conferences. He is currently the Editor-in-Chief of Aerospace Science and Technology and serves on the IEEE Control Systems Society Conference Editorial Board. In addition, he is an Associate Editor for several leading journals, including Automatica, IEEE Transactions on Automatic Control, IEEE Transactions on Neural Networks and Learning Systems, IEEE Transactions on Systems, Man, and Cybernetics: Systems, IEEE Transactions on Artificial Intelligence, Neural Networks, IEEE Open Journal of the Computer Society, and the Journal of Optimization Theory and Applications. Previously, he has served as Guest Senior Editor for special issues of IEEE Transactions on Automation Science and Engineering, IEEE Transactions on Neural Networks and Learning Systems, IEEE Transactions on Industrial Informatics, IEEE Transactions on Intelligent Transportation Systems, and the IEEE Open Journal of Control Systems. Dr. Vamvoudakis is a registered Professional Engineer (Electrical/Computer Engineering), a member of the Technical Chamber of Greece, an Associate Fellow of AIAA, and a Senior Member of IEEE.