Program

Liquid liquid phase separation is a widespread phenomenon in living cells, that takes advantage of spontaneous condensation of certain proteins and RNA species to spatially and dynamically organize a variety of molecules and reactions. Taking inspiration from nature, our group is developing methods to build liquid condensates and patterns using artificial DNA and RNA molecules, that are relevant to the synthesis of self-regulated, autonomous biomaterials. I will present our recent results on controlling the rate of condensation and dissolution of artificial DNA and RNA motifs using chemical reactions. I will also discuss mathematical models that support and guide our experiments by capturing specific and non-specific interactions among the motifs.

Elisa Franco received her B.S. and M.S. (Laurea Degree) in Power Systems Engineering from the University of Trieste (Italy) in 2002, summa cum laude. In 2007, she received her Ph. D. in Automation from the same institution. In 2011, she completed her second Ph. D. at the California Institute of Technology, Pasadena, in Control and Dynamical Systems. Dr. Franco is currently an Associate Professor of Mechanical and Aerospace Engineering and Bioengineering in the Molecular Biology Institute at UCLA. Prior to UCLA, she served as an Assistant Professor in the Mechanical Engineering Department at the University of California, Riverside (2011-2018.)

Cities are central to today's sustainability challenges, including public health and safety, environmental impacts, and equity and access. At the same time, cities are becoming more like robots, with increasingly pervasive sensing and new forms of actuation. From a lens of robotics, machine learning, control, and transportation engineering, there is a once-in-a-generation opportunity to learn effective interventions to move the needle on long-standing societal challenges. However, urban settings are massively multi-agent, safety-critical yet impossible to model perfectly, and highly varied. This talk focuses on our recent work in addressing the scalability of learning methods in distributed environments, providing assurances for massive robotic operations, and examining the robustness of learning methods to environmental diversity.

Cathy Wu is an Assistant Professor at MIT in LIDS, CEE, and IDSS. She holds a Ph.D. from UC Berkeley, and B.S. and M.Eng. from MIT, all in EECS, and completed a Postdoc at Microsoft Research. She studies the role of machine learning and computation in the design of future mobility systems. Her interests include reinforcement learning, autonomy, multi-agent dynamical systems, urban systems, and combinatorial optimization. Her work has been acknowledged by several awards, including the 2019 IEEE ITSS Best Ph.D. Dissertation Award, 2019 Microsoft Location Summit Hall of Fame, 2018 Milton Pikarsky Memorial Dissertation Award, the 2016 IEEE ITSC Best Paper Award, and numerous fellowships, and appeared in the press, including Wired and Science Magazine.

During a cognitively demanding task or at rest, the brain exhibits a rich repertoire of large-scale functional patterns. These patterns are a measure of the coherence among the neural activities in different brain areas, reflect different cognitive functions, and can also be used as biomarkers for different psychiatric and neurological disorders. For example, while patterns of transient and partial coherence have been observed in healthy individuals, increased coherence in neural systems is often associated with degenerative diseases including Parkinson’s and Huntington's disease, and epilepsy. In this talk, I will discuss methods to model, analyze, and control functional patterns in oscillatory brain networks. I will start by modeling the rhythmic activity of a neural system as the output of a network of diffusively coupled heterogeneous oscillators, and use different synchronization notions as a proxy for functional patterns. I will derive conditions for the emergence of cluster synchronization, where independent groups of synchronized oscillators coexist in a network, and compare such conditions against empirical brain data. Finally, I will present a method to enforce desired synchronization patterns through minimally invasive and localized changes to the network structure, validate some of our findings using a well-accepted neurovascular model, and discuss future research directions.

Fabio Pasqualetti is a Professor of Mechanical Engineering at the University of California, Riverside. He completed a Doctor of Philosophy degree in Mechanical Engineering at the University of California, Santa Barbara, in 2012, a Laurea Magistrale degree (M.Sc. equivalent) in Automation Engineering at the University of Pisa, Italy, in 2007, and a Laurea degree (B.Sc. equivalent) in Computer Engineering at the University of Pisa, Italy, in 2004. He is the recipient of the 2017 Young Investigator Award from the Army Research Office and the 2019 Young Investigator Research Award from the Air Force Office of Scientific Research. His articles received the 2016 TCNS Outstanding Paper Award, the 2019 ACC Best Student Paper Award, the 2020 Control Systems Letters Outstanding Paper Award, the 2020 Roberto Tempo Best CDC Paper Award, and the 2021 O. Hugo Schuck Best Paper Award. His main research interests are in the areas of network systems, computational neuroscience, and machine learning.

Many optimization problems relevant to the design and operation of electric power systems are inherently nonlinear due to the AC power flow equations that model the relationships between the voltages and the power flows in power grids. The nonlinearity of the power flow equations results in a variety of algorithmic and theoretical challenges, including non-convex feasible spaces for optimization problems containing these equations. This presentation describes four categories of methods for addressing these challenges: 1) local optimization, 2) approximation, 3) convex relaxation, and 4) convex restriction. Local optimization methods search for an operating point that is superior to all nearby points. The practical applicability of local optimization methods has been demonstrated via results from the US Department of Energy’s Grid Optimization Competition, which compared algorithms for solving large-scale security-constrained AC optimal power flow problems. Approximations, convex relaxations, and convex restrictions simplify the power flow equations to obtain more tractable convex representations that are useful in a variety of applications. This presentation describes recent research on these methods and discusses several relevant applications.

Daniel Molzahn is an Assistant Professor in the School of Electrical and Computer Engineering and a Fellow of the Strategic Energy Institute at the Georgia Institute of Technology. He also holds an appointment as a computational engineer in the Energy Systems Division at Argonne National Laboratory, where he was previously a member of the research staff. He was a Dow Postdoctoral Fellow in Sustainability at the University of Michigan and received the B.S., M.S., and Ph.D. degrees in electrical engineering and the Masters of Public Affairs degree from the University of Wisconsin–Madison, where he was a National Science Foundation Graduate Research Fellow. He received the IEEE Power and Energy Society’s 2021 Outstanding Young Engineer Award for contributions to the theory and practical application of nonlinear optimization algorithms for electric power systems.

Many of today’s most promising technological systems involve very large numbers of autonomous agents that influence each other and make strategic decisions within a network structure. Examples include opinion dynamics and product adoption in social networks, economic exchange and international trade in financial networks, as well as security investment decisions in cyber-physical systems. While traditional tools for the analysis of these systems assumed that a social planner has full knowledge of the underlying network of interactions, when we turn to very large populations three issues emerge. First, collecting data about the exact network of interactions becomes very expensive or not at all possible because of privacy concerns. Second, methods for designing optimal interventions that rely on the exact network structure typically do not scale well with the population size. Third, individual preferences of the agents may depend on unknown parameters that need to be inferred from observations of agents' actions. To obviate these issues, in this talk I will consider a framework in which the social planner uses probabilistic instead of exact information about agent’s interactions. I will introduce the tool of “graphon games” as a way to formally describe strategic interactions for infinite populations and I will illustrate how this tool can be exploited to tractably design interventions and infer unknown parameters for broad classes of networked multi-agent systems involving large but finite populations.

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 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.

A rising challenge in control of large-scale control systems such as the electricity and the transportation networks is to address autonomous decision making of interacting agents, i.e. the subsystems, with local objectives while ensuring global system safety and performance. In this setting, a Nash equilibrium is a stable solution outcome in the sense that no agent finds it profitable to unilaterally deviate from her decision. Due to geographic distance, privacy concerns or simply the scale of these systems, each agent can only base her decision on local measurements. Hence, a fundamental question is: do agents learn to play a Nash equilibrium strategy based only on local information? I will discuss conditions under which we have an affirmative answer to this question and will present algorithms that achieve this learning task.

Maryam Kamgarpour holds a Doctor of Philosophy in Engineering from the University of California, Berkeley and a Bachelor of Applied Science from University of Waterloo, Canada. Her research is on safe decision-making and control under uncertainty, game theory and mechanism design, mixed integer and stochastic optimization and control. Her theoretical research is motivated by control challenges arising in intelligent transportation networks, robotics, power grid systems, financial markets and healthcare. She is the recipient of NASA High Potential Individual Award, NASA Excellence in Publication Award, the European Union (ERC) Starting Grant and NSERC Discovery Accelerator Grant.

A team of multiple cooperative autonomous robotics can drastically improve the benefits of robotic deployments by overcoming the physical limits of individual robots. Such potential has driven considerable research efforts worldwide, and this talk aims to address research developments related with planning and coordination in multi-robot systems necessary for full exploitation of their capabilities. In particular, it will discuss approaches to efficiently process distributed local measurements for exploring a large environments, to generate physically feasible trajectories for cooperative operation of multiple agents, and to coordinate multiple robots for effective utilization of limited available resources will be discussed. Experimental demonstrations using ground and flying robots including aerial exploration and autonomous transportation will also be presented.

H. Jin Kim is a Professor in Aerospace Engineering at Seoul National University. received MSc and PhD degrees in Mechanical Engineering from the University of California, Berkeley and BS in Mechanical Engineering from Korea Advanced Institute of Science and Technology (KAIST), Korea. Her research is on navigation, control and path planning of autonomous robotic systems, such as ground robots, autonomous vehicles, and flying robots. She has served as an Associate Editor for several journals and conferences including IEEE Transactions on Robotics, Mechatronics, an International Journal of IFAC, and IEEE Conference on Robotics and Automation, and an Editor for International Journal of Control, Automation, and Systems. She has been selected as one of the leading researchers for 100 future technologies in Korea and a member of National Academy of Engineering of Korea.

There has been a large body of work studying how agents can learn communication protocols in decentralized settings, using their actions to communicate information. Surprisingly little work has studied how this can be prevented, yet this is a crucial prerequisite from a human-AI and AI-safety point of view. The standard problem setting in Dec-POMDPs is self-play, where the goal is to find a set of policies that play optimally together. Policies learned through self-play may adopt arbitrary conventions and implicitly rely on multi-step reasoning based on fragile assumptions about other agents' actions and thus fail when paired with humans or independently trained agents at test time. To address this, we present off-belief learning (OBL). At each timestep OBL agents follow a policy pi_1 that is optimized assuming past actions were taken by a given, fixed policy (pi_0, but assuming that future actions will be taken by pi_1. When pi_0 is uniform random, OBL converges to an optimal policy that does not rely on inferences based on other agents' behavior (an optimal grounded policy). OBL can be iterated in a hierarchy, where the optimal policy from one level becomes the input to the next, thereby introducing multi-level cognitive reasoning in a controlled manner. Unlike existing approaches, which may converge to any equilibrium policy, OBL converges to a unique policy, making it suitable for zero-shot coordination (ZSC). OBL can be scaled to high-dimensional settings with a fictitious transition mechanism and shows strong performance in both a toy-setting and the benchmark human-AI & ZSC problem Hanabi.

Jakob Foerster started as an Associate Professor at the department of engineering science at the University of Oxford in the fall of 2021. During his PhD at Oxford he helped bring deep multi-agent reinforcement learning to the forefront of AI research and interned at Google Brain, OpenAI, and DeepMind. After his PhD he worked as a research scientist at Facebook AI Research in California, where he continued doing foundational work. He was the lead organizer of the first Emergent Communication workshop at NeurIPS in 2017, which he has helped organize ever since and was awarded a prestigious CIFAR AI chair in 2019. His past work addresses how AI agents can learn to cooperate and communicate with other agents, most recently he has been developing and addressing the zero-shot coordination problem setting, a crucial step towards human-AI coordination. His work has been cited over 5000 times, with an h-index of 29 (Google Scholar page).

In this talk, I will present a game-theoretic model of strategic, online news dissemination on Twitter-like social networks. Agents are endowed with subjective, heterogeneous priors on some unobservable real-valued state of the world. At the beginning, a small fraction of agents observes a piece of news with certain credibility. The agents who receive the news, decide to share or not with their followers based on whether the news will persuade their followers to move their beliefs closer to the agents’. We then characterize agents’ sharing decision, which leads to an endogenous SI process. We characterize the size of endogenous news spread at the equilibrium. We show that low-credibility news can trigger a larger sharing cascade than news with higher credibility, when the society has diverse priors and the network is highly connected . Furthermore, we investigate the role of polarization in priors and show that increased polarization in a population prompts more sharing of lower credibility news which results in wider spread than fully credible news. Next , I will discuss a model of news subscription where news intermediaries have ideological biases, and a population with heterogeneous priors is deciding to which news intermediary to subscribe. We show how the equilibrium of this game and the model rationalizes homophily for individuals at the ideological extremes. Joint work with Chi-Chia Hsu, Amir Ajorlou, and Muhamet Yildiz.

The advent of shared-economy and smartphones made on-demand transportation services possible, which created additional opportunities, but also more complexity to urban mobility. Companies that offer these services are called Transportation Network Companies (TNCs) due to their internet-based nature. Although ride-sourcing is the most notorious service TNCs provide, little is known about to what degree its operations can interfere in traffic conditions, while replacing other transportation modes, or when a large number of idle vehicles is cruising for passengers. We experimentally analyze the efficiency of TNCs using taxi trip data from a Chinese megacity and an agent-based simulation with a trip-based MFD model for determining the speed. We investigate the effect of expanding fleet sizes for TNCs, passengers’ inclination towards sharing rides, and strategies to alleviate urban congestion. We observe that, although a larger fleet size reduces waiting time, it also intensifies congestion, which, in turn, prolongs the total travel time. Asymmetry in demand also creates imbalances in the spatial distribution of these vehicles. Parking management and rebalancing strategies based on principles of hierarchical and distributed control can prevent idle vehicles from cruising without assigned passengers in congested areas and improving the service quality.

Nikolas Geroliminis is a Professor at EPFL and the head of the Urban Transport Systems Laboratory (LUTS). Before joining EPFL he was an Assistant Professor at the University of Minnesota. He has a diploma in Civil Engineering from the National Technical University of Athens (NTUA) and a MSc and Ph.D. in civil engineering from University of California, Berkeley. His research interests focus primarily on urban transportation systems, traffic flow theory and control, public transportation and on-demand transport, car sharing, Optimization, MFDs and Large Scale Networks. He is a recipient of the ERC Starting Grant METAFERW: Modeling and controlling traffic congestion and propagation in large-scale urban multimodal networks. Among his recent initiatives is the creation of an open-science large-scale dataset of naturalistic urban trajectories of half a million vehicles that have been collected by one-of-a-kind experiment by a swarm of drones ( https://open-traffic.epfl.ch). He currently serves as Editor-In-Chief of Transportation Research part C: Emerging Technologies journal.

The electric power system is currently transitioning from a centralized system with bulk power generation and inflexible demand to a more distributed infrastructure due to the increased penetration of distributed generation resources, storage and flexible demand. The renewable generation resources such as wind and PV increase overall uncertainty and variability in the electric energy supply but the control capabilities of the distributed resources also provide great potential to keep the system balanced and to ensure secure and stable system operation. However, in order to leverage this potential, new approaches to operating the system are required. Possibilities include distributed and decentralized control approaches, local intelligence based on learning techniques and more. The change in system dynamics when moving from large synchronous machines to distributed inverter based generation thereby plays an important role and needs to be considered when redesigning system operation. This talk will give an overview over the challenges that need to be addressed and present a few examples of what role control can play to overcome these challenges.

Prof. Gabriela Hug received the M.Sc. degree in information technology and electrical engineering in 2004 and the Ph.D. degree in electric power systems in 2008 from ETH Zurich. She also received the diploma in Higher Education Teaching in Electrical Engineering in 2007 from the same institution. After her PhD, she worked in the Special Studies Group of Hydro One in Toronto, Canada and from 2009 - 2015 she was an Assistant Professor at Carnegie Mellon University in Pittsburgh, USA. Currently, she is Professor in the Power Systems Laboratory at ETH Zurich. Her research is dedicated to control and optimization of electric power systems. She has received a number of awards for her work including the US NSF CAREER Award, the IEEE PES Outstanding Young Engineer Award and the ALEA Award.

Cyber-Physical Systems have been instrumental in bringing together talented researchers from different domains to focus their attention on developing a paradigm capable of addressing modern real-world system design issues, as separation of concerns does not constitute a realistic assumption, due to the close interplay of sensing, communication, computing and decision making. As a result, system-level research has become more relevant and impactful. In this talk I will provide a personal view of the progress made in CPS since inception and provide a perspective on where the field is headed. In particular I will focus on the issue of guaranteeing resilience and trustworthiness while leveraging modern data driven methods in the presence of large uncertainties and adversarial actions.

Bruno Sinopoli is the Das Family Distinguished Professor at Washington University in St. Louis, where he is also the founding director of the center for Trustworthy AI in Cyber-Physical Systems and chair of the Electrical and Systems Engineering Department. He received the Dr. Eng. degree from the University of Padova in 1998 and his M.S. and Ph.D. in Electrical Engineering from the University of California at Berkeley, in 2003 and 2005 respectively. After a postdoctoral position at Stanford University, Dr. Sinopoli was member of the faculty at Carnegie Mellon University from 2007 to 2019, where he was a professor in the Department of Electrical and Computer Engineering with courtesy appointments in Mechanical Engineering and in the Robotics Institute and co-director of the Smart Infrastructure Institute. His research interests include modeling, analysis and design of Resilient Cyber-Physical Systems with applications to Smart Interdependent Infrastructures Systems, such as Energy and Transportation, Internet of Things and control of computing systems.

Facebook down for hours; Nearly half of Japan without Internet; thousands of airline passengers grounded; 600+ emergency calls not received. These are just few recent illustrations of the consequences of network misconfigurations. Such errors are caused by a fundamental gap: distilling the high-level requirements network operators have in mind into complex, low-level device configurations. In 2021, network operators are still trying to bridge this gap manually and, by doing so, end up making mistakes. In this talk, I will speak about our quest to solve this problem, describes how it relates to control, our successes thus far, and the hurdles we are (still) trying to overcome. Among others, I will speak about our project of building a formally-verified, declarative network controller where operators simply tell the network what to do, not how to do it.

Laurent Vanbever is an associate professor at ETH Zürich where he started as an assistant professor in 2015. Before that, Laurent was a postdoctoral research associate at Princeton University where he worked with Jennifer Rexford. He obtained his PhD degree in Computer Science from the University of Louvain in 2012. His research focuses on making large network infrastructures more manageable, scalable and, secure. Laurent has won several awards for his work including: the SIGCOMM Rising Star Award, two SIGCOMM best paper awards, the NSDI community award, six IETF/IRTF Applied Networking Research Prizes, and an ERC Starting Grant.