CyDiSy Topics and Speakers

Abstract: 

Digital Twins have the potential to reduce cost, improve quality, and expand capabilities in manufacturing systems.  As computing and networking technologies improve, the massive amounts of data being collected on manufacturing plant floors can be leveraged through digital twins to create useful information and advise human operators on recommended actions.  Even with the huge amount of data available, data quality remains an important challenge.  Standards for Digital Twins are emerging, and there are opportunities to create different types of Digital Twins that can best utilize the data that exits.  In this talk, we will present a requirements framework for Digital Twins in the manufacturing domain, including the important properties of re-usability, interoperability, interchangeability, extensibility and maintainability. Several examples of digital twins that we have created, in collaboration with our industry partners, will be presented, covering multiple application domains.  Future challenges and opportunities in the area will also be discussed, including how automated learning approaches can be utilized to update and enhance digital twin solutions. 

Bio:  

Dawn M. Tilbury is the inaugural Ronald D. and Regina C. McNeil Department Chair of Robotics at the University of Michigan, and the Herrick Professor of Engineering. She received the B.S. degree in Electrical Engineering from the University of Minnesota, and the M.S. and Ph.D. degrees in Electrical Engineering and Computer Sciences from the University of California, Berkeley.  Her research interests lie broadly in the area of control systems, including applications to robotics and manufacturing systems.  From 2017 to 2021, she was the Assistant Director for Engineering at the National Science Foundation, where she oversaw a federal budget of nearly $1 billion annually, while maintaining her position at the University of Michigan. She has published more than 200 articles in refereed journals and conference proceedings.  She is a Fellow of IEEE, a Fellow of ASME, and a Life Member of SWE. 

Abstract 

Electricity generation is a significant contributor to global CO2 emissions, and the datacenter industry is expected to reach anywhere from 3 to 13% of global electricity demand by 2030. Datacenters have the potential to facilitate grid decarbonization in a manner different from isolated power loads, and Google has set on a mission to increase and scale carbon awareness via new technology solutions. We present an overview and enhancements of the previously announced Google’s Carbon-Intelligent Computing System, and discuss technical challenges associated with increasing and harnessing temporal and spatial flexibility of diverse workloads running in Google datacenters. Furthermore, we share some insights from building systems that leverage different types of workload flexibility to effectively increase resource efficiency and reduce environmental impact, while meeting infrastructure and application SLOs. Finally, we outline the results from our investigations of load shaping strategies aimed not only to reduce grid-level emissions, but contribute to energy systems’ more resilient, robust and cost-efficient decarbonization. 

Biography 

Ana Radovanovic has been a research scientist at Google since early 2008, after she earned her PhD Degree in Electrical Engineering from Columbia University (2005) and worked for 3 years as a Research Staff Member in the Mathematical Sciences Department at IBM TJ Watson Research Center. For the last 10 years, Ana Radovanovic has focused all her research efforts at Google on building innovative technologies and business models with two goals in mind: (i) to deliver more reliable, affordable and clean electricity to everyone in the world, and (ii) to help Google become a thought leader in decarbonizing the electricity grid. Nowadays, Ana is widely recognized as a technical lead and research entrepreneur. She is a Senior Staff Research Scientist, serving as a Technical Lead for Energy Analytics and Carbon Aware Computing at Google. 

Abstract

This talk considers the problem of resilient distributed multi-agent optimization for cyberphysical systems in the presence of malicious or non-cooperative agents. It is assumed that stochastic values of trust between agents are available which allows agents to learn their trustworthy neighbors simultaneously with performing updates to minimize their own local objective functions. The development of this trustworthy computational model combines the tools from statistical learning and distributed consensus-based optimization. Specifically, we derive a unified mathematical framework to characterize convergence, deviation of the consensus from the true consensus value, and expected convergence rate, when there 

exists additional information of trust between agents. We show that under certain conditions on the stochastic trust values and consensus protocol: 1) almost sure convergence to a common limit value is possible even when malicious agents constitute more than half of the network, 2) the deviation of the converged limit, from the nominal no attack case, i.e., the true consensus value, can be bounded with probability that approaches 1 exponentially, and 3) correct classification of malicious and legitimate agents can be attained in finite time almost surely. Further, the expected 

convergence rate decays exponentially with the quality of the trust observations between agents. We then combine this trust-learning model within a distributed gradient-based method for  solving a multi-agent optimization problem and characterize its performance. 

Biography

Angelia Nedich has a Ph.D. from Moscow State University, Moscow, Russia, in Computational Mathematics and Mathematical Physics (1994), and a Ph.D. from Massachusetts Institute of Technology, Cambridge, USA in Electrical and Computer Science Engineering (2002). She has worked as a senior engineer in BAE Systems North America, Advanced Information Technology Division at Burlington, MA. Currently, she is a faculty member of the school of Electrical, Computer and Energy Engineering at Arizona State University at Tempe. Prior to joining Arizona State University, she has been a Willard Scholar faculty member at the University of Illinois at Urbana-Champaign. She is a recipient (jointly with her co-authors) of the Best Paper Award at the Winter Simulation Conference 2013 and the Best Paper Award at the International Symposium on Modeling and Optimization in Mobile, Ad Hoc and Wireless Networks (WiOpt) 2015.  Her general research interest is in optimization, large scale complex systems dynamics, variational inequalities and games. 

Abstract

This talk will explore the synergistic history of using computers to solve optimization problems and optimization methods to design computing systems. Beginning with the role of Dantzig and linear programming, I’ll trace how the formulation of optimization as a field influenced the design of early computers. Next, I’ll describe how the resulting computers enabled and shaped developments in optimization and control. I’ll then discuss how the resulting coalescence of optimization defined how we design computers and cyber-physical systems more generally. I’ll close with a speculative discussion of where these ideas might take us as Moore’s Law scaling winds down.

Biography

Benjamin Recht is a Professor in the Department of Electrical Engineering and Computer Sciences at the University of California, Berkeley. His research has focused on applying mathematical optimization and statistics to problems in data analysis and machine learning. He is currently studying histories, methods, and theories of scientific validity and experimental design.

Abstract

Reinforcement learning (RL) has had tremendous successes in many artificial intelligence applications. Many of the forefront applications of RL involve multiple agents, e.g., playing chess and Go games, autonomous driving, and robotics. Unfortunately, classical RL framework is inappropriate for multi-agent learning as it assumes an agent’s environment is stationary and does not take into account the adaptive nature of opponent behavior. In this talk, I focus on stochastic games for multi-agent reinforcement learning in dynamic environments and develop independent learning dynamics for stochastic games: each agent is myopic and chooses best-response type actions to other agents’ strategies independently, meaning without any coordination with her opponents. There has been limited progress on developing convergent best-response type independent learning dynamics for stochastic games. I will present our recently proposed independent learning dynamics that guarantee convergence in zero-sum stochastic games. We then focus on the minimal information setting where agents do not observe opponent’s actions, but only observe the payoff they receive at each round. We present payoff-based and independent learning dynamics for such settings and provide finite-time guarantees using a novel coupled Lyapunov drift approach.

Biography 

Asu Ozdaglar is the Mathworks Professor of Electrical Engineering and Computer Science (EECS) at the Massachusetts Institute of Technology (MIT). She is the department head of EECS and deputy dean of academics of the Schwarzman College of Computing at MIT. Her research expertise includes optimization, machine learning, economics, and networks. Her recent research focuses on designing incentives and algorithms for data-driven online systems with many diverse human-machine participants. She has investigated issues of data ownership and markets, spread of misinformation on social media, economic and financial contagion, and social learning.
Professor Ozdaglar is the recipient of a Microsoft fellowship, the MIT Graduate Student Council Teaching award, the NSF Career award, the 2008 Donald P. Eckman award of the American Automatic Control Council, the 2014 Spira teaching award, and Keithley, Distinguished School of Engineering and Mathworks professorships. She is an IEEE fellow, IFAC fellow, and was selected as an invited speaker at the International Congress of Mathematicians. She received her Ph.D. degree in electrical engineering and computer science from MIT in 2003.

Abstract

Reports of cyber-attacks, such as Stuxnet, have shown their devastating consequences on digitally controlled systems supporting modern societies, and shed light on their modus operandi: First learn sensitive information about the system, then tamper the visible information so the attack is undetected, and meanwhile have significant impact on the physical system. Securing control systems against such complex attacks requires a systematic and thorough approach. In the first part of the talk, we provide an overview of recent work on secure networked control systems centered on a risk management framework. In the second part of the talk, we focus on specific sensor attack scenarios and mitigation strategies currently being investigated in our research group. We show that an attacker with access to the sensor channel can perfectly estimate a linear controller’s state without error, and thus violate the operator’s privacy, if and only if the controller has no unstable poles. An advanced attacker may exploit such a breach of confidentiality to design stealthy false data injection attacks (violations of sensor data integrity) resulting in large physical impact on the controlled plant. We illustrate some of the results using lab experiments, and discuss moving target defense mitigation strategies. 

Abstract

Modern socio-technical systems involve a large number of nodes or agents interacting in heterogeneous ways. It is clear that interventions aimed at improving the performance or resilience of these systems should exploit information about the underlying network of interactions, yet most systems of interest are of very large dimension introducing several challenges for the analysis and design of such large scale networked systems. In this talk, I will illustrate a general methodology to overcome these challenges by i) introducing the concept of a graphon system, which captures the dynamic behavior of an infinite population of agents via the graph limit concept of graphons, and ii) developing a convergence theory of large networked systems to graphon systems. I will illustrate the applicability of this framework by considering two key dynamical processes modeling contagion and synchronization behavior via the linear threshold and Kuramoto model, respectively. 

Biography

Francesca Parise joined the School of Electrical and Computer Engineering at Cornell University as an assistant professor in July 2020. Before then, she was a postdoctoral researcher at the Laboratory for Information and Decision Systems at MIT. She defended her PhD at the Automatic Control Laboratory, ETH Zurich, Switzerland in 2016 and she received the B.Sc. and M.Sc. degrees in Information and Automation Engineering in 2010 and 2012, from the University of Padova, Italy, where she simultaneously attended the Galilean School of Excellence. Francesca’s research focuses on identification, analysis and control of multi-agent systems, with application to transportation, energy, social and economic networks. 

Francesca was recognized as an EECS rising star in 2017, she is a finalist for the 2022 ISSNAF Young Investigator Award and is the recipient of the Guglielmo Marin Award from the “Istituto Veneto di Scienze, Lettere ed Arti”, the SNSF Early Postdoc Fellowship, the SNSF Advanced Postdoc Fellowship and the ETH Medal for her doctoral work.

Abstract

In direct data-driven control the design of control policies is reduced to the solution of data-dependent convex programs. The majority of the available results focuses on linear systems, while a few extensions are obtainable for special classes of nonlinear systems. Unsurprisingly, deriving solutions for general nonlinear systems is much harder. In this talk, we discuss a method to design feedback control laws from data that render a nonlinear system dominantly linear.   

Biography

Claudio De Persis is a professor with the Engineering and Technology Institute, University of Groningen, the Netherlands, since 2011. He received his Ph.D. degree in Engineering from the University of Rome "la Sapienza", Italy, in 2000, after which he was a postdoctoral researcher with Washington University in St. Louis, MO and Yale University, CT. Before joining the University of Groningen, he was with the University of Rome "la Sapienza" and Twente University, Enschede, the Netherlands. His research interest is in automatic control and his current research focuses on the use of data for control design. 

Abstract

In cyber-physical systems (CPS), the interaction between the control, communication, and computation (CCC) elements and the physical processes of the plant is crucial for safe and high-performance closed-loop behaviour. Unfortunately, the traditional approach of "separating of concerns" for the design of CCC components is no longer feasible or far from optimal for many CPS applications. This is especially true for many networked control systems (NCS), where the overall performance depends crucially on both the communication networks and the control systems. In this talk, we will review a range of results that analyse and design NCS using different hybrid modelling formalisms. We will start by examining "communication in control," where we take the perspective that communication networks and their limitations introduce time-varying sampling effects, such as varying transmission intervals, delays, packet losses, and possible denial-of-service attacks in the closed-loop dynamics. Additionally, scheduling protocols can be present that dictate which sensor or control nodes can access the communication network at transmission times. We will quantitatively investigate the impact of these network-induced artefacts on the stability and performance of the controlled systems. In contrast, in the "control in communication" part, we consider not only the design of control systems but also "control" when sensor and controller nodes are allowed to communicate. Specifically, we will focus on event-triggered control, where transmissions are generated smartly based on the system's state or output, ensuring desirable stability and performance properties while saving communication resources by only communicating when necessary. In fact, we will present cases where event-triggered control strictly outperforms the traditional paradigm of time-triggered periodic control, and discuss relevant applications such as cooperative adaptive cruise control (CACC) strategies for vehicle platooning. We close the talk by discussing the connections between the "communication in control" and "control in communication" parts, identifying open problems, and glimpse into the future of this fascinating research area. 

Biography

Maurice Heemels received M.Sc. (mathematics) and Ph.D. (EE, control theory) degrees (summa cum laude) from the Eindhoven University of Technology (TU/e) in 1995 and 1999, respectively. From 2000 to 2004, he was with the Electrical Engineering Department, TU/e, as an assistant professor, and from 2004 to 2006 with the Embedded Systems Institute (ESI) as a Research Fellow, during which he also was at Oce [Canon]. Since 2006, he has been with the Department of Mechanical Engineering, TU/e, where he is currently a Full Professor and Vice-Dean. He held visiting professor positions at ETH, Switzerland (2001), UCSB, USA (2008) and University of Lorraine, France (2020). He is a Fellow of the IEEE and IFAC, and the chair of the IFAC Technical Committee on Networked Systems (2017-2023). He served/s on the editorial boards of Automatica,  Nonlinear Analysis: Hybrid Systems (NAHS), Annual Reviews in Control, and IEEE Transactions on Automatic Control, and is the Editor-in-Chief of NAHS as of 2023. He was a recipient of a personal VICI grant awarded by NWO (Dutch Research Council) and an ERC Advanced Grant (2021). He was the recipient of the 2019 IEEE L-CSS Outstanding Paper Award and was elected for the IEEE-CSS Board of Governors (2021-2023).   His current research includes hybrid and cyber-physical systems, networked and event-triggered control systems and model predictive control. 

Abstract

In this talk, we will introduce a new approach to building neural networks and nonlinear dynamical models with built-in guarantees of stability, robustness and other certified behavioural properties. We will trace the connections between convex parameterizations from robust control to so-called “direct” parameterizations, i.e. smooth & unconstrained parameterisations of all models that satisfy prescribed conditions. These direct parameterizations enable learning of robust static and dynamic models via simple first-order methods, without any auxiliary constraints or projections. We will explore some applications in certifiably-robust image classification, physics-informed learning of contracting nonlinear observers, and robust reinforcement learning for nonlinear partially-observed systems. 

Biography

Ian R. Manchester received the B.E. and Ph.D. degrees in electrical engineering from the University of New South Wales, Sydney, NSW, Australia, in 2002 and 2006, respectively. From 2006-2009 he was a post-doctoral researcher at Umea University, Sweden, and from 2009-2012 he was a Research scientist at the Massachusetts Institute of Technology. Since 2012 he has been a faculty member at the University of Sydney, Australia, where he is currently Professor of Mechatronic Engineering, Director of the Australian Centre for Robotics, and Director of the Australian Robotic Inspection & Asset Management Hub. His current research interests are in algorithms for control, estimation, and learning of nonlinear dynamical systems, with applications in robotics, robust machine learning, biomedical engineering, and smart networked systems. 

Abstract

In his talk, Arno Schlueter will discuss the role of automation and controls for future-proof buildings. To mitigate building-related greenhouse gas emissions while maintaining human comfort and adapting to a changing climate, active and passive building systems are becoming increasingly sophisticated and networked. Building automation and controls plays a critical role, but efforts associated with and interactions with the building users become key. Examples from research in Singapore and Zurich will be used to discuss different approaches to building controls for future-proof buildings, the challenges and the respective learnings obtained from operating different living labs.

Biography

Arno Schlueter holds a degree (Dipl.Ing.Architektur) from the Technical University of Karlsruhe, a postgraduate degree in CAAD and a PhD in building systems from ETH Zurich. In 2010, he was appointed Assistant Professor and in 2014, Full Professor of Architecture and Building Systems (A/S) at the Institute of Technology in Architecture (ITA), ETH Zurich. Since 2013, he is also a Principal Investigator at the Singapore-ETH Future Cities Lab (FCL). He currently is the Head of the Institute of Technology in Architecture of ETH Zurich and a member of the management board of the ETH Energy Science Centre.

In his research, he and his interdisciplinary team focus on sustainable building systems, new adaptive components and their synergetic integration into architecture and urban design using computational approaches for modelling, analysis, and control. In 2009, he co-founded the design and engineering office KEOTO.ch, where he is part of the management board. The work of the A/S Group has been published in scientific journals, magazines and books and has won international competitions and prizes.