ROCOND'18 Plenary Talk: Easy, hard or convex?: the role of sparsity and structure in robust control and identification.
Speaker:Mario Sznaier, College of Engineering, Northeastern University, USA
Abstract:Arguably, one of the hardest challenges faced now by the systems community stems from the exponential explosion in the availability of data, fueled by recent advances in sensing and actuation capabilities. Simply stated, classical techniques are ill equipped to handle very large volumes of (heterogeneous) data, due to poor scaling properties, and to impose the structural constraints required to implement ubiquitous sensing and control. For example, the powerful Linear Matrix Inequality framework developed in the past 20 years and associated semidefinite program based methods have proven very successful in providing global solutions to many control and identification problems. However, in may cases these methods break down when considering problems involving just a few hundred data points. On the other hand, several in-principle non-convex problems (e.g identification and robust control of classes of switched systems) can be efficiently solved in cases involving large amounts of data. Thus the traditional convex/non-convex dichotomy may fail to completely capture the intrinsic difficulty of some problems. The goal of this talk is to explore how this "curse of dimensionality" can be potentially overcome by exploiting the twin "blessings” of self-similarity (high degree of spatio-temporal correlation in the data) and inherent underlying sparsity, and to answer the question of "what is Big Data in systems theory?". While these ideas have already been recently used in machine learning (for instance in the context of dimensionality reduction and variable selection), they have hitherto not been fully exploited in systems theory. By appealing to a deep connection to semi-algebraic optimization, rank minimization and matrix completion we will show that, in the context of systems theory, the limiting factor is given by the "memory" of the system rather than the size of the data itself, and discuss the implications of this fact. These concepts will be illustrated by examining examples of "easy" and "hard" problems, including identification and control of hybrid systems and (in)validation of switched models. We will conclude the talk by exploring the connection between hybrid systems identification, information extraction, and machine learning, and point out to new research directions in systems theory and in machine learning motivated by these problems.
Plenary video on IFAC YouTube channel: https://youtu.be/8O-lEKB3Mpg
Speaker: José C. Geromel, School of Electrical and Computer Engineering, UNICAMP, Brazil.
Abstract: Bellman's Principle of Optimality is the main tool to formulate sampled- data control design problems as convex programming ones for LTI and Markov Jump Linear Systems. Stability and performance conditions are determined and expressed through differential linear matrix inequalities (DLMIs). This important fact opens the possibility to deal with plant parameters uncertainty in convex domains. Optimal control and guaranteed cost control problems in the context of H2 and H ∞ norms are considered. Design problems to be numerically solved are handled upon constraints are converted to LMIs. The limit case characterized by arbitrarily small sampling period is discussed. Examples are presented to illustrate the theoretical results.Plenary video on IFAC YouTube channel: https://youtu.be/TMriZfhvQS4
Biography: José C. Geromel received the Docteur d'Etat degree from University Paul Sabatier - LAAS / CNRS, Toulouse, France, in 1979. He joined the School of Electrical and Computer Engineering / UNICAMP in 1975, where he is professor of Control Systems Analysis and Design. He was awarded, in 1994 and 2014, with the Zeferino Vaz Award for his teaching and research activities at UNICAMP and, in 2007, with the Scopus Award jointly promoted by Elsevier and CAPES / Brazil. Since 1991 he has been a Fellow of the CNPq - Brazilian Council for Research Development. From 1998 to 2002 he was the Dean for Graduate Studies of UNICAMP. Since 1998 he is a member of the Brazilian Academy of Science. In 1999 he has been named Chevalier dans l'Ordre des Palmes Academiques by the French Minister of National Education. In 2010 he received the Docteur Honoris Causa degree from University Paul Sabatier, Toulouse, France and he has been named a member of the Brazilian Order of Scientific Merit by the President of the Federative Republic of Brazil. In 2011 he has been named Distinguished Lecturer by the IEEE Control Systems Society.
Speaker: Karl H. Johansson, KTH Royal Institute of Technology
Abstract: Freight transportation on roads is of utmost importance for our society. It accounts for a significant amount of all energy consumption and greenhouse gas emissions. In this talk, we will discuss the potential future of road goods transportation and how it can be made more robust and efficient, from the automation of individual long-haulage trucks to the optimisation of fleet management and logistics. Such an integrated cyber-physical transportation system benefits from having trucks travelling together in vehicle platoons. From the reduced air drag, platooning trucks travelling close together can save more than 10% of their fuel consumption. In addition, by automating the driving, it is possible to change driver regulations and thereby increase the efficiency even more. Control and optimization problems on various level of this transportation system will be presented. It will be argued that a system architecture utilising vehicle-to-vehicle and vehicle-to-infrastructure communication enable robust and safe control of individual trucks as well as optimised vehicle fleet collaborations and new market opportunities. Extensive experiments done on European highways will illustrate system performance and safety requirements. The presentation will be based on joint work over the last ten years with collaborators at KTH and at the truck manufacturer Scania.Plenary video on IFAC YouTube channel: https://youtu.be/u41SFl-5Ys4
Short Biography Karl Henrik Johansson is Director of the Stockholm Strategic Research Area ICT The Next Generation and Professor at the School of Electrical Engineering and Computer Science, KTH Royal Institute of Technology. He received MSc and PhD degrees from Lund University. He has held visiting positions at UC Berkeley, Caltech, NTU, HKUST Institute of Advanced Studies, and NTNU. His research interests are in networked control systems, cyber-physical systems, and applications in transportation, energy, and automation. He is a member of the IEEE Control Systems Society Board of Governors, the IFAC Executive Board, and the European Control Association Council. He has received several best paper awards and other distinctions. He is a Distinguished Professor with the Swedish Research Council and a Wallenberg Scholar. He has received the Future Research Leader Award from the Swedish Foundation for Strategic Research and the triennial Young Author Prize from IFAC. He is Fellow of the IEEE and the Royal Swedish Academy of Engineering Sciences, and he is IEEE Distinguished Lecturer.
LPVS'18 Plenary Talk: Robust Modelling, Analysis and Control in Aerospace: from LFT and H-infinity to LPV
Speaker:Andrés Marcos University of Bristol (UK)
Abstract: In this talk several aerospace industrial study cases are used to illustrate the validity of robust modeling, analysis and control techniques. The results showcase the fundamental importance of the linear fractional transformation (LFT) and linear parameter varying (LPV) robust modeling paradigms. These models enable the application of robust analyses methods (e.g. structured singular value and integral quadratic constraints) that complement, and serve as a bridge for, the classical linear and nonlinear simulation analyses used by industry. Furthermore, their concurrent use with robust control synthesis approaches (from structured H-infinity to LPV) allows to formalize a modeling, analysis and design framework that facilitates the inherently difficult performance versus robustness trade-off across wide system variations. As a conclusion of the talk it is shown that such a framework using LPV control design is (at least for the case of launcher systems in atmospheric phase) more methodological, capable and transparent than the more advanced and complex adaptive approaches currently in favour in academia and industry.
Short Biography: Andrés Marcos is from Córdoba, Spain. He received his Aerospace Engineering B.Sc. from St. Louis University (USA) in 1997, and his M.Sc. and Ph.D. degrees in 2000 and 2004 respectively from the University of Minnesota (USA) in the group of prof. Gary Balas. From 2006 to August 2013 he worked at Deimos-Space S.L.U. (Spain) as lead R&D control engineer for projects from the European Space Agency (ESA) and the European Commission (EC-FP7). In October 2013, he joined the Aerospace Engineering department at the University of Bristol (UK), were he formed and leads the Technology for Aerospace Control (TASC) lab, http://www.tasc-group.com/.
The aim of TASC is the study and application of robust techniques to aeronautical and space systems. TASC technical focus is on: flight control (FCS); fault detection, identification and reconfiguration (FDIR); flight control Verification and Validation (V&V); and research on linear fractional transformation (LFT) and linear parameter varying (LPV) techniques for modeling, design and analysis.
Speaker Ricardo S. Sánchez Peña
Abstract: Here we present a methodology to obtain low order LPV models focused on controller design for Diabetes Mellitus type 1, also called the Artificial Pancreas. The data to compute these models comes from the distributed version of the well-known UVa/Padova simulator. The LPV dynamics take into account the time-varying and nonlinear nature of the problem, as well as the subcutaneous-intravenous delays. The model is tuned to each patient by means of his/her clinical information. The resulting dynamics are presented as a delayed affine LPV model, which makes it amenable to LPV design methods. The efficiency of this model is measured in terms of quadratic errors and by the $\nu$-gap metric. An enhancement of these LPV models, which includes the intra-patient variations, is also discussed. An application example is presented, describing the first Latin America clinical trial, which used the ARG control algorithm.Plenary video on IFAC YouTube channel: https://youtu.be/vzGPFp6Sia8
Short Biography: He received the Electronic Engineer degree from the University of Buenos Aires (UBA, 1978), and the M.Sc. and Ph.D. from the California Institute of Technology (1986, 1988), both in Electrical Engineering. In Argentina he worked between 1977 and 2004 in CITEFA, CNEA and the space agencies CNIE and CONAE. He collaborated with NASA, the German (DLR) and Brazilian (CTA/INPE) space agencies. He was (Plenary) Full Professor at UBA (1989-2004), ICREA Senior Researcher at the UPC (2005-2009, Barcelona) and visiting Prof./Researcher at several Universities in the USA and the EU. He consulted for ZonaTech (USA), Alstom-Ecotecnia (Spain) and STI and VENG (Argentina). He published 4 books and more than 160 journal and conference papers. He received awards from NASA, IEEE and the National Academy of Exact, Physical and Natural Sciences of Argentina. He is Director of the Research and PhD Program at the Buenos Aires Institute of Technology(ITBA) and a CONICET Principal Investigator. He has applied Identification and Control techniques to acoustical, mechanical, aero and astronautical engineering, and currently to type 1 diabetes and neurobiology.