Over the past two decades, we have witnessed an unprecedented growth in sensing, communications, computation, and robotic actuation, which can drastically change the way our society collects and processes information. The sensor network revolution has created the possibility of exploring and interacting with the environment in ways not possible before. The recent progress in robotics and actuation has further created the possibility of unmanned autonomous vehicles helping us achieve tasks that are otherwise hazardous or impossible. The growing importance of multi-agent cooperative systems has also been heavily acknowledged by the research community, and several work has appeared addressing different aspects of these systems. However, we are still quite far from fully understanding how to design a multi-agent system such that all the issues of sensing, communication, navigation, and resource/system constraints are addressed in a general framework that is as independent of a specific scenario as possible. We are not even close to fundamentally understanding how to design a multi-agent system in terms of the homogeneity and diversity of its team members. Furthermore, due to the multi-disciplinary nature of these systems, research has also been conducted in several different communities in parallel. For instance, in the area of distributed decision making, the control community has mainly focused on the theoretical foundations of consensus while the robotics community has developed algorithmic approaches to deal with the complexity of the problem. In the area of connectivity maintenance, the control community has focused on the graph theoretical approaches, while the networking community has addressed routing and multi-hop design issues on mobile systems. As such, after several years of extensive work in this area, it is the right time to pause and ask about the important problems that are still open and need to be solved to fundamentally move the field forward. This is the main goal of this proposed workshop. By bringing together experts that have extensively worked in different aspects of multi-agent systems, we hope to have a clear understanding of the road ahead in realizing the full vision of these systems. Note that the purpose of the workshop is not to present a collection of recent results but rather to explicitly highlight what we still, as a community, do not know!
Back to top
Distributed decision making
Coordination and group agreements
Navigation and path planning
Sensing and coverage
Role of game theory and auctions in multi-agent system design
Role of machine learning in multi-agent system design
Co-optimization of sensing, communication and navigation
Understanding heterogeneity of multi-agent systems
Human machine interaction
Optimality and guarantees of performance
Capacity of a multi-agent system
Back to top
Dr. Francesco Bullo, Professor, University of California Santa Barbara and Dr. Florian Dörfler, Professor, University of California Los Angeles
Abstract: We consider a set of selected problems in the areas of energy and social networks. As first problem set, we consider the two fundamental stability problems in interconnected power grids: synchronization on the active power side and voltage collapse on the reactive power side. We highlight a few recent advances based on multi-agent modeling and analysis, and we present a list of open analysis and distributed control problems. As second problem set, we consider co-evolution problems in social networks. In particular, we are interested in the interplay between opinion dynamics and evolving influence networks. We highlight a few recent advances and present a list of open problems in modeling and analysis of co-evolving networks.
Bio: Francesco Bullo (S'95-M'99-SM'03-F'10) is a Professor with the Mechanical Engineering Department at the University of California, Santa Barbara. He received a Laurea in Electrical Engineering from the University of Padova in 1994, and a Ph.D. in Control and Dynamical Systems from the California Institute of Technology in 1999. From 1998 to 2004, he was affiliated with the University of Illinois, Urbana-Champaign. His students' papers were finalists for the Best Student Paper Award at the IEEE Conference on Decision and Control (2002, 2005, 2007), and the American Control Conference (2005, 2006, 2010). He is the coauthor of the books "Geometric Control of Mechanical Systems" (Springer, 2004) and "Distributed Control of Robotic Networks" (Princeton, 2009). He is a recipient of the 2003 ONR Young Investigator Award, the 2008 IEEE CSM Outstanding Paper Award and the 2011 AACC Hugo Schuck Best Paper Award. His main research interest is multi-agent networks with applications to robotic coordination, power systems and distributed computing.
Bio:Florian Dörfler is an Assistant Professor in the Department of Electrical Engineering at the University of California Los Angeles, and he is affiliated with the Center for Nonlinear Studies at the Los Alamos National Laboratories. He received a Ph.D. degree in Mechanical Engineering from the University of California at Santa Barbara in 2013, and a Diplom degree in Engineering Cybernetics from the University of Stuttgart in 2008. His primary research interests are centered around distributed control, complex networks, and cyber-physical systems with applications to smart power grids and robotic coordination. He is recipient of the the 2010 ACC Student Best Paper Award, the 2011 O. Hugo Schuck Best Paper Award, and the 2014 Automatica Best Paper Award. As a co-advisor and a co-author, he has been a finalist for the ECC 2013 Best Student Paper Award.
Dr. Mehran Mesbahi, Professor, University of Washington
Abstract: In this talk I motivate and describe a set of problems in the area multi-agent coordination and control that have a distinct combinatorial character. These problems relate to state-dependent dynamic networks, combinatorial analysis of the notion of network performance, and network-centric notions of learning and adaptation.
Bio: Mehran Mesbahi received his Ph.D. from USC in 1996. He was a member of the Guidance, Navigation, and Analysis group at JPL from 1996-2000 and an Assistant Professor of Aerospace Engineering and Mechanics at University of Minnesota from 2002-2002. He is currently a Professor of Aeronautics and Astronautics, an Adjunct Professor of Mathematics, and the Executive Director of the Joint Center for Aerospace Technology Innovation at the University of Washington in Seattle. He was the recipient of NSF CAREER Award in 2001, NASA Space Act Award in 2004, UW Distinguished Teaching Award in 2005, and UW College of Engineering Innovator Award in 2008. His research interests are distributed and networked aerospace systems, systems and control theory, and engineering applications of optimization and combinatorics.
Dr. Jeff Shamma, Professor, Georgia Institute of Technology
Abstract: One approach to designing distributed algorithms for multi-agent systems is to view the interacting agents from the perspective of game theory. Game theory is a framework for modeling and analyzing interacting decision makers. By definition game theory is clearly relevant to multi-agent systems, and there has been ongoing work in recent years to apply game theory to distributed coordination in multi-agent systems. Indeed, even when game theory is not explicitly invoked, algorithms such as distributed averaging for consensus can be reinterpreted from game theoretic perspectives. The basic elements of a game are the set of players, their choices, and their preferences, which are typically characterized by a utility function. These elements for multi-agent systems constitute design degrees of freedom for a system planner, as opposed to being an outcome of a modeling procedure. Accordingly, the application of game theory as a design tool for engineered systems is contrary to the more traditional role of game theory as a model is social sciences. This talk revisits game theoretic approaches to multi-agent system design through a discussion of its potential and limitations as a design tool and a presentation of selected open problems.
Bio: Jeff Shamma is the Julian T. Hightower Chair in Systems & Control in the School of Electrical and Computer Engineering at the Georgia Institute of Technology. Jeff received a BS in Mechanical Engineering from Georgia Tech in 1983 and a PhD in Systems Science and Engineering from the Massachusetts Institute of Technology in 1988. Prior to returning to Georgia Tech in 2007, he held faculty positions at the University of Minnesota, University of Texas-Austin, and University of California-Los Angeles. Jeff is a recipient of the NSF Young Investigator Award (1992) and the American Automatic Control Council Donald P. Eckman Award (1996), and a Fellow of the IEEE (2006). He previously served on the Air Force Scientific Advisory Board (2008-2011) and is currently an Associate Editor for the IEEE Transactions on Cybernetics (2009-present) and Games (2012-present) and a Senior Editor for the newly formed IEEE Transactions on Control of Network Systems.
Dr. Angela schoellig, Professor, University of Toronto
Abstract: Traditionally, motion planning and control algorithms (for example, for controlling the motion of a robot) have been designed based on a priori knowledge about the system such as dynamic models of the system, maps of the environment and information about the desired task. This approach has enabled successful (robot) operations in predictable environments; for example, in manufacturing plants or warehouses. However, to achieve reliable operations in unknown, changing and generally uncontrolled environments, we must enable agents to acquire knowledge during operation and to adapt their behavior accordingly. The next question that arises is whether the knowledge collected by one agent is beneficial for another agent. In this talk, I will first show motivating examples that demonstrate that learning is important for achieving high-performance agent behavior. You will see videos of rovers traversing rough terrain and flying vehicles racing through a slalom course. I will then highlight the questions that arise when considering multi-agent learning and will describe first approaches towards answering those questions. Ultimately, I want to convey that multi-agent learning is a new and exciting field of study – just imagine all agents uploading their knowledge to the “cloud”, benefiting from fast cloud computing and making the essence of their knowledge available to other agents.
Bio: Angela Schoellig is an Assistant Professor at the University of Toronto Institute for Aerospace Studies (UTIAS). She conducts research at the interface of robotics, controls and learning. Her goal is to enhance the performance and autonomy of robots by enabling them to learn from past experiments and from each other. Angela has worked with aerial vehicles for the past six years. She has been developing planning, control and learning algorithms for high-performance robot motions. You can watch her vehicles perform slalom races and flight dances at www.youtube.com/user/angelaschoe. Angela received her Ph.D. from ETH Zurich (supervised by Prof. Raffaello D’Andrea), and holds both an M.Sc. in Engineering Science and Mechanics from the Georgia Institute of Technology and a Masters degree in Engineering Cybernetics from the University of Stuttgart. Her Ph.D. was awarded the ETH Medal and the Dimitris N. Chorafas Foundation Award. She was finalist of the 2008 IEEE Fellowship in Robotics and Automation, which supports prospective leaders in this field. Her past research has been published in journals such as Autonomous Robots and the IEEE Robotics & Automation Magazine, and has received coverage worldwide in mainstream TV, print and online media. More information about her research is available at: www.schoellig.name.
Dr. Magnus Egerstedt, Professor, Georgia Institute of Technology
Abstract: While we, as a community, have made great inroads towards cataloguing and characterizing a number of situations and scenarios where heterogeneous teams are useful, our understanding of the role of heterogeneity itself is quite limited. While we do talk about such concepts as air plus ground coordination, a mothership with smaller robots, and related ideas, results in these areas have a flavor of case studies rather being systematic scientific studies. We do not even have clear definitions of heterogeneity let alone answer such as questions as: When is a heterogeneous solution better than a homogeneous one? How do heterogeneous solutions emerge/evolve? What role does heterogeneity play in nature, especially as the cognitive and physical abilities of individual agents increase? In this talk, these issues will be discussed and a possible road forward will be introduced.
Bio: Magnus Egerstedt is the Schlumberger Professor in the School of Electrical and Computer Engineering at the Georgia Institute of Technology, where he also serves as Associate Chair for Research and External Affairs. He received the M.S. degree in Engineering Physics and the Ph.D. degree in Applied Mathematics from the Royal Institute of Technology, Stockholm, Sweden, and the B.A. degree in Philosophy from Stockholm University. Dr. Egerstedt conducts research in the areas of control theory and robotics, with particular focus on control and coordination of complex networks, such as multi-robot systems, mobile sensor networks, and cyber-physical systems. He is the director of the Georgia Robotics and Intelligent Systems Laboratory (GRITS Lab), a Fellow of the IEEE, and a recipient of the ECE/GT Outstanding Junior Faculty Member Award, the HKN Outstanding Teacher Award, the Alumn of the Year Award from the Royal Institute of Technology, and the U.S. National Science Foundation CAREER Award.
Dr. Yasamin Mostofi, Professor, University of California Santa Barbara
Abstract: In this talk, I will discuss open problems at the intersection of robotics and communications for the energy-aware operation of multi-agent systems. Several work has focused on the individual optimization of navigation and communication issues in mobile robotic networks. Under resource constraints, however, it becomes considerably important to take an integrative approach and co-optimize the sensing, communications and navigation aspects of these problems. A related aspect is the dynamics of information generation, gathering and exchange, which is dictated by the design as well as the maximum capacity of the system. However, a proper characterization of the relationship between entropy rate of information generation in the environment, and capacity of sensing/communication of the network is lacking. This is important as it helps us decide on the feasibility of a design as well as its maximum performance beforehand. In this talk, I will focus on these problems and will discuss a plan for moving forward.
Bio: Yasamin Mostofi received the BS degree in electrical engineering from Sharif University of Technology, Tehran, Iran, in 1997, and the MS and PhD degrees in the area of wireless communications from Stanford University, California, in 1999 and 2004, respectively. She is currently an associate professor in the Department of Electrical and Computer Engineering at the University of California Santa Barbara. Dr. Mostofi is the recipient of the Presidential Early Career Award for Scientists and Engineers (PECASE), the National Science Foundation (NSF) CAREER award, and IEEE 2012 Outstanding Engineer Award of Region 6. She also received the Bellcore fellow-advisor award from Stanford Center for Telecommunications in 1999 and the 2008-2009 Electrical and Computer Engineering Distinguished Researcher Award from the University of New Mexico. She has served on the IEEE Control Systems Society conference editorial board 2008-2013 and is currently an associate editor for the IEEE Transactions on Control of Network Systems. She is a senior member of the IEEE.
Back to top
coffee break 10:00-10:15am
Lunch Break 11:45am-1:30pm
Coffee Break 3:00-3:30pm
Back to top
We expect the workshop to be of interest to researchers working on different aspects of multi-agent systems such as coordination, sensing, distributed decision making, navigation, connectivity control, and energy related issues. We find the workshop in particular useful for young researchers such as students as it provides new directions for future research.
Back to top