You are on page 1of 27

European Embedded Control Institute

Graduate School on Control

Independent Modules one 21 hours module per week (3 ECTS)
Deadline for ADVANCE REGISTRATION to each module: 20/12/2012
Locations: Supelec (Paris), Istanbul (Turkey), LAquila (Italy), Belgrade (Serbia)
M1 Randomized Algorithms for Systems and 14/01/2013 18/01/2013 Control: Theory and Applications M2 Uncertain Optimization via 21/01/2013 25/01/2013 Sample-Based Approaches M3 Model Predictive Control 28/01/2013 01/02/2013 M4 The Transverse Function Control Approach 04/02/2013 08/02/2013 for Highly Nonlinear Systems M5 Design and analysis tools for physical 11/02/2013 15/02/2013 control systems M6 Normal Forms for Nonlinear Control 18/02/2013 22/02/2013 Systems and Their Applications M7 Decentralized and Distributed Control 25/02/2013 01/03/2013 M8 Modeling and Control of Automotive and 04/03/2013 08/03/2013 Aerospace Engines and Powerplants M9 Stability and Control of Time-delay 11/03/2013 15/03/2013 Systems M10 Recent Advances of Sliding Mode Control 11/03/2013 15/03/2013 M11 - BELGRADE Control of Nonlinear Delay Systems 11/03/2013 15/03/2013 and PDEs Verification and Correct-by-Construction M12 - BELGRADE Synthesis of Control Protocols for 18/03/2013 22/03/2013 Networked Systems M13 Input saturation: control design and 18/03/2013 22/03/2013 anti-windup M14 25/03/2013 - 29/03/2013 M15 25/03/2013 29/03/2013 M16 08/04/2013 12/04/2013 Traffic modeling and estimation at the age of smartphones Model Predictive Control About Nonlinear Digital Control Roberto Tempo, CNR-IEIIT, Politecnico di Torino, Italy Fabrizio Dabbene, CNR-IEIIT, Politecnico di Torino, Italy Marco C. Campi, University of Brescia, Italy Simone Garatti, Politecnico di Milano DEI, Italy

Eduado F. Camacho, University of Sevilla, Spain

Claude Samson, INRIA, France Pascal Morin, ISIR, Universit Pierre et Marie Curie, France Antonio Lora, CNRS L2S, Gif-sur-Yvette, France Elena Panteley, CNRS L2S, Gif-sur-Yvette, France Witold Respondek, INSA Rouen, France Giancarlo Ferrari-Trecate, University of Pavia, Italy Marcello Farina, Politecnico di Milano, Italy Ilya Kolmanovsky, University of Michigan, USA Wim Michiels, K.U. Leuven, Belgium Silviu-Iulian Niculescu, CNRS L2S, Gif-sur-Yvette, France Vadim I. Utkin, The Ohio State University, USA Miroslav Krstic, University of California, San Diego, USA Richard Murray, California Institute of Technology, USA Ufuk Topcu, California Institute of Technology, USA Nok Wongpiromsarn, Singapore-MIT Alliance Research &Tech Sophie Tarbouriech, CNRS LAAS, Toulouse, France Luca Zaccarian, CNRS LAAS, Toulouse, France Alexandre M. Bayen, University of California, Berkeley, USA Dan Work, University of Illinois at Urbana-Champaign, USA Christian Claudel, University of Sci. and Tech. Thuwal, KSA Jan Maciejowski, University of Cambridge, UK

Dorothe Normand-Cyrot, CNRS L2S, Gif-sur-Yvette, France Salvatore Monaco, University of Roma La Sapienza, Italy W.P.M.H. Heemels, Eindhoven Univ. of Tech., Netherlands M17 Event-triggered and Self-triggered Control Karl-Henrik Johansson, Royal Institute of Tech. Sweden 22/04/2013 26/04/2013 Paulo Tabuada, University of California at Los Angeles, USA M18 - ISTANBUL Stochastic Control with Contemporary Roger W. Brockett, Harvard School of Eng. Applied Sc., USA 22/04/2013 26/04/2013 Methods and Applications Maria Domenica Di Benedetto, University of LAquila, Italy M19 - ISTANBUL Symbolic control design of Giordano Pola, University of LAquila, Italy 29/04/2013 03/05/2013 Cyber-Physical systems Alessandro Borri, IASI-CNR, Rome, Italy M20 Alessandro Astolfi, Imperial College, UK Nonlinear and Adaptive Control 13/05/2013 17/05/2013 Romeo Ortega, CNRS L2S, Gif-sur-Yvette, France M21 Distributed Control A. Stephen Morse, Yale University, USA 13/05/2013 17/05/2013 M22 Extremum Seeking Control: Dragan Nesic, University of Melbourne, Australia 20/05/2013 24/05/2013 Analysis and Design M23 Robust Hybrid Control Systems Ricardo Sanfelice, University of Arizona, USA 20/05/2013 24/05/2013 M24 LAQUILA Optimality, Stabilization, and Feedback Francis Clarke, Universit Claude Bernard Lyon 1, France 20/05/2013 24/05/2013 in Nonlinear Control M25- LAQUILA Emmanuel Witrant, Univ. Joseph Fourier, GIPSA, Grenoble, Modeling and estimation for control 27/05/2013 31/05/2013 France M26 Switched Systems and Control Daniel M. Liberzon, University of Illinois, USA 27/05/2013 31/05/2013

(*) A module will open

only if a sufficient number of registrations are received before the advance registration deadline: 20/12/2012

European Embedded Control Institute

M1 14/01/2013 18/01/2013

Randomized Algorithms for Systems and Control: Theory and Applications

Roberto Tempo
CNR-IEIIT, Politecnico di Torino, Italy

Fabrizio Dabbene
CNR-IEIIT, Politecnico di Torino, Italy

Abstract of the course In this course, we provide a perspective of the area of randomization for systems and control, and study several topics which include the computation of the sample complexity and the connections with statistical learning theory. In particular, we address system's analysis and design using sequential and non-sequential randomized methods, and analyze advantages and disadvantages of these approaches. In the second part, we show how randomization is successfully used in several applications within and outside engineering. We present an overview of these methods for aerospace and automotive control, hard disk drives, systems biology, congestion control of networks, quantized, switched and embedded systems, multi-agent consensus. Particular emphasis is given on the computation of PageRank in Google, web aggregation techniques, and control design of UAVs. The course is based on the book by R. Tempo, G. Calafiore, F. Dabbene, Randomized Algorithms for Analysis and Control of Uncertain Systems with Applications, 2nd edition, Springer-Verlag, London, 2012. Topics: - Uncertain systems - Probabilistic methods for analysis - Monte Carlo and Quasi-Monte Carlo algorithms - Random sampling techniques - Probabilistic methods for control design - Probability inequalities and statistical learning theory

European Embedded Control Institute

M2 21/01/2013 25/01/2013

Uncertain Optimization via Sample-Based Approaches

Marco C. Campi
Department of Information Engineering University of Brescia, Italy

Simone Garatti
Dipartimento di Elettronica ed Informazione Politecnico di Milano, Italy

Abstract of the course: Optimization problems involving uncertainty are ubiquitous, and emerge in diverse domains ranging from control to allocation, from planning to finance. In this course, we shall introduce the student to sample-based approaches where uncertainty is described by means of a finite number of samples, or scenarios, coming from the infinite set of possible uncertainty outcomes. Sample-based approaches represent a viable solution methodology in a variety of optimization problems involving uncertainty. Samples can as well be observations, and this covers data-based approaches in learning and identification. A particular emphasis in the course will be given to the scenario approach. The presentation will be gradual to allow an in-depth understanding of the fundamental concepts. Special attention will be given to a precise mathematical formulation of the problems and to a detailed presentation of the ensuing results. Practical examples will illustrate the ideas.


- Uncertain optimization - Monte-Carlo sampling - Scenario approach - Applications to various domains - Discussion of open problems that offer an opportunity for research

European Embedded Control Institute

M3 28/01/2013 01/02/2013

Model Predictive Control

Abstract of the course: Model Predictive Control (MPC) has developed considerably in the last decades both in industry and in academia. Although MPC is considered to be a mature discipline, the field has still many open problems and attracts the attention of many researchers. This courses provides an extensive review concerning the theoretical and practical aspects of predictive controllers. It describes the most commonly used MPC strategies, showing both the theoretical properties and their practical Eduardo F. Camacho implementation issues. As part of the course the Dept. System Engineering and Automatica students will program and simulate different MPC University of Seville , Spain structures. Special focus is made in the control of a real solar energy plant that will serve as an application example of the different techniques reviewed in the course. The course is designed around the text book: E. F. Camacho and C. Bordons, Model Predictive Control, 2nd edition, Springer, 2004 Prerequisites: Undergraduate-level knowledge of differential equations and control systems. Topics: 1. Introduction to MPC, process models, disturbance models, prediction equations. 2. MPC used in industry: FIR and step response based MPC. DMC. 3. MPC used in academy: GPC and State Space based MPC. 4. MPC of multivariable processes, dead time problems, choosing the control horizons, MPC and transmission zeros. Practical aspects for implementing multivariable MPC. 5. MPC and constraints: Handling constraints, QP and LP algorithms. Solving the constrained MPC, multi-parametric methods. Constrained and stability in MPC. 6. Nonlinear MPC, parametric models, local based function models, optimization methods. 7. Stability and robustness in MPC: Stability guaranteed MPCs, robust stability for MPC, robust constraint satisfaction, Min-max MPC. 8. Open issues: multi-objective MPC, MPC of hybrid systems, the tracking problem in MPC, distributed and hierarchical MPC, cooperative MPC. 9. MPC application to a solar power plant: plant models, MPC and intraday market, MPC and RTO: dynamical optimal set point determination, MPC for set point tracking. Choosing the appropriate models and horizon for each control level.

European Embedded Control Institute

M5 04/02/2013 08/02/2013

The Transverse Function Control Approach for Highly Nonlinear Systems

Claude Samson
INRIA, France

Pascal Morin
UPMC, France

Abstract of the course: The course in an introduction to the Transverse Function approach recently developed by P. Morin and C. Samson to control nonlinear systems that are locally controllable at equilibria but whose linear approximation is not. Such systems are sometimes referred to as "critical" systems. The non-existence of asymptotical stabilizers in the form of continuous pure-state feedback controllers, as pointed out by a Brockett's theorem for a large subclass of critical systems, calls for the development of control solutions that depart from "classical" nonlinear control theory. An important motivation for the control engineer arises from the fact that many physical systems can be modeled as critical systems. Such is the case, for instance, of nonholonomic mechanical systems (like most mobile vehicles on wheels, ranging from common car-like vehicles to ondulatory wheeled-snake robots) and of many underactuated vehicles (like ships, submarines, hovercrafts, blimps). Beyond these theoretical aspects, an important motivation for the control engineer also arises from the fact that many physical systems can be modeled as critical systems. Such is the case, for instance, of nonholonomic mechanical systems (like most mobile vehicles on wheels, ranging from common car-like vehicles to ondulatory wheeled-snake robots) and of many underactuated vehicles (like ships, submarines, hovercrafts, blimps). Asynchronous electrical motors also belong to this category. Topics include: Controllability and stabilization properties of critical systems Homogeneous approximation of critical controllable systems Lie group invariance properties of homogeneous drftless systems Definition, existence and calculation of Transverse Functions Feedback control design by the Transverse function approach Application to nonholonomic or underactuated systems

European Embedded Control Institute

M5 11/02/2013 15/02/2013

Design and Analysis Tools for Physical Control Systems

Antonio Loria
CNRS - France loria@lss.supelec

Elena Panteley
CNRS - France

Abstract of the course: Departing from the premise that the world is nonlinear, dynamic and deterministic, physics laws are omnipresent to study the behaviour of systems and their interactions with their environment. Regardless of the engineering discipline, if Automatic Control is the spine of technology, Lyapunov stability theory lays at the foundations of model-based control and qualitative analysis.

This course covers a selected number of tools, useful to analyse the stability and performance of controlled systems in which physical properties and engineering intuition are the main steering reins of the control designer. The presentation is streamlined by particular systems structures such as in the case of Model Reference Adaptive Control, cascaded systems, passive interconnections For pedagogical reasons, particular attention is put into case-studies stemming from control of robotic systems, consensus, formation control, electromechanical systems, synchronization, etc.

- Stability analysis of time-varying systems, adaptive control, output feedback - control, robust control, observer design, separation principle ...

European Embedded Control Institute

M6 18/02/2013 22/02/2013

Normal Forms for Nonlinear Control Systems and Their Applications

Abstract of the course: The aim of this course is to present a fairly complete list of normal forms for various classes of nonlinear control systems. Such forms have been obtained during the last 30 years for various purposes: classification, stabilization, tracking, motion planning, observation etc. We will attempt to present them in a systematic way, by providing normal forms, necessary and sufficient conditions for equivalence to them, and (whenever they exist) algorithmic procedures for obtaining them. We will show usefulness of the presented forms in various nonlinear control problems: linearization, flatness, stabilization, output and trajectory tracking, and nonlinear observers. Outline:

Witold Respondek
INSA de Rouen, France

1. Feedback and state equivalence. 2. Feedback linearizable systems. - Globally feedback linearizable systems. - Partial feedback linearization. 3. Special classes of control systems. - Systems on R2 - Locally simple systems. 4. Triangular forms. - Lower triangular forms and feedback linearizability. - p-normal forms. - Upper triangular forms and feedforward systems. - Linearizable feedforward systems 5. Formal feedback and formal normal forms. - General systems. - Feedforward systems. 6. Flatness, dynamic feedback, and normal forms for subclasses of flat systems - Normal forms for driftless systems: chained forms. - Normal forms versus search for flat outputs. 7. Nonlinear control systems with observations. - Local normal forms. - Global normal forms

European Embedded Control Institute

M7 25/02/2013 01/03/2013

Decentralized and Distributed Control

Giancarlo Ferrari-Trecate
Dipartimento di Ingegneria Industriale e dellInformazione Universita' degli Studi di Pavia, Italy ferrari/welcome.html

Marcello Farina
Dipartimento di Elettronica ed Informazione Politecnico di Milano, Italy

Abstract of the course: Advances in technology and telecommunications are steadily broadening the range and size of systems that can be controlled. Examples that bring new challenges for control engineering are smart grids, that are perceived as the future of power generation, and networks of sensors and actuators, that enable the monitoring and control of processes spread over large geographical areas. As an alternative to centralized regulators, that seldom make sense for large-scale systems, decentralized and distributed approaches to control have been developed since the seventies. Particular attention has been recently given to distributed control architectures based on model predictive control that are capable to cope with physical constraints. The first part of the course will focus on classical results on stability analysis of large-scale systems, decentralized control and decentralized controllability issues. Then, distributed control design methods will be covered. In the last part of the course, more emphasis will be given to recent advances in distributed control strategies based on optimization and receding horizon control.
Topics: - Introduction to large-scale systems and multivariable control - Decentralized control architectures - Stability analysis of large-scale systems - Decentralized controllability issues and design of decentralized control systems - Design of distributed control systems

European Embedded Control Institute

M8 04/03/2013 08/03/2013

Modeling and Control of Automotive and Aerospace Engines and Powerplants

Abstract of the course: With increasing stringency of fuel efficiency and emissions requirements, opportunities emerge to improve engine performance through model-based control. This course will provide an introduction to modeling, estimation and control problems for engines and powerplants in automotive applications, and a briefer perspective on related problems in aerospace applications. The use of control-theory based and model-based approaches will be emphasized. Approaches to handling constraints in engines using reference governors and model predictive control will be discussed in detail. The topics covered include techniques for developing engine control-oriented models, control and estimation problems for naturally aspirated and turbocharged gasoline engines, and modeling and control of diesel engines. Topics of engine-transmission coordination and energymanagement for Hybrid Electric Vehicles will also be covered. Related modeling, control and constraint handling problems for aircraft gas turbine and internal combustion engines, and for hybrid aircraft powerplant will also be discussed.

IIya Kolmanovsky
Department of Aerospace Engineering University, of Michigan, USA people/faculty/kolmanovsky/index.html

Topics: 1. Basic principles and techniques of engine control-oriented modeling 2. Modeling, estimation and control of naturally aspirated gasoline engines 3. Modeling and control problems for turbocharged gasoline engines 4. Modeling and control problems for diesel engines 5. Constraint handling in automotive engines based on reference governors and model predictive control 6. Engine-transmission coordination 7. Hybrid Electric Vehicle energy management 8. Gas turbine engine modeling and control problems 9. Limit protection for gas turbine engines 10. Hybrid powerplant energy management in aircraft applications 11. Perspective and discussion on control challenges and opportunities for advanced and future engines

European Embedded Control Institute

M9 11/03/2013 15/03/2013

Stability and Control of Time-delay Systems

Wim Michiels
Department of Computer Science KU Leuven, Belgium

Silviu Niculescu
Lanoratoire des Signaux et Systmes CNRS - Suplec , France

Abstract of the course: Time-delays are important components of many systems from engineering, economics and the life sciences, due to the fact that the transfer of material, energy and information is mostly not instantaneous. They appear, for instance, as computation and communication lags, they model transport phenomena and heredity and they arise as feedback delays in control loops. The aim of this course is to describe fundamental properties of systems subjected to time-delays and to present an overview of methods and techniques for the analysis and control design. The focus lies on systems described by functional differential equations and on frequency-domain techniques, grounded in numerical linear algebra (e.g., eigenvalue computations, matrix distance problems) and optimization. Several examples (from chemical to mechanical engineering, from tele-operation to high-speed networks, from biological systems to population dynamics) complete the presentation. Topics: Theory: Classification and representation of time-delay systems Definition and properties of solutions of delay differential equations Spectral properties of linear time-delay systems Computational methods: Stability determining eigenvalues Stability domains in parameter spaces Robustness and performance measures Controller synthesis via eigenvalue optimization Control design: Fundamental limitations induced by delays Fixed-order optimal H-2 and H-infinity controllers Prediction based controllers Using delays as controller parameters

European Embedded Control Institute

M10 11/03/2013 15/03/2013

Recent Advances of Sliding Mode Control

Abstract of the course: I - Introduction. Mathematical Tools. Design Principles. The principal design idea of sliding mode control implies selection of discontinuous control enforcing the state trajectories to the pre-selected manifold with a reduced order motion equations and desired properties of this motion. Mathematical methods for analysis of differential equations with discontinuous right-hand parts are surveyed along with their applications for designing feedback control systems.

Vadim I. Utkin
Department of Electrical Engineering The Ohio State University, USA

II Higher order sliding mode control The question of interest whether similar effect can be reached for the cases with relative degree greater than one, or when control input is a continuous state function. Then the range of applications of sliding mode control will be increased. In numerous publications different design methods for sliding mode control as a continuous state function were offered and the authors referred to their methods as high order sliding mode control. The design methods will be discussed in the presentation except for the cases when high order sliding modes can be easily interpreted in terms of the conventional sliding modes (or first order sliding modes). The main attention will be paid to the so-called twisting and supertwisting algorithms. III Chattering suppression Alternative methods of chattering suppression the main obstacle for sliding mode control implementation - are discussed in this part. As a rule chattering is caused by unmodelled dynamics. The first recipe is application of asymptotic observers. They serve as a bypass for high frequency component in control and as a result the unmodelled dynamics are not excited. However under uncertainty conditions the conventional observers can not be used for chattering suppression. Another way to reduce chattering implies state-dependent magnitude of discontinuous control, since the chattering amplitude is a monotonously increasing function of the discontinuity magnitude. The methodology is not applicable for widely used electronic power converters with constant magnitude of a discontinuous output. For these systems the efficient tool to suppress chattering is harmonic cancellation principle. IV - Applications Applications of the sliding mode control and observation methodology along with chattering suppression are demonstrated for electric machines, power converters and automotive engines.

European Embedded Control Institute

M11 - BELGRADE 12/03/2013 16/03/2013

Control of Nonlinear Delay Systems and PDEs

Abstract of the course:
In the 1990s, the recursive backstepping design enabled the creation of adaptive and robust control algorithms for nonlinear systems with nonlinearities of unlimited growth and with uncertainties that are not matched by control. Taking the backstepping recursion to the continuous limit provides a design methodology for boundary control of PDEs and for some key classes of delay systems. Contrary to standard PDE control that mimics LQR for finite-dimensional systems and yields virtually intractable operator Riccati equations, backstepping feedback laws come with explicit gain formulas. This course, mostly based on the instructors book Boundary Control of PDEs: A Course on Backstepping Designs (SIAM, 2008), teaches how to derive such formulas for specific classes of PDE systems.

Miroslav Krstic
Department of Mechanical & Aero. Eng. University of California, San Diego, USA

The explicit feedback laws allow the design of previously inconceivable parameter-adaptive controllers for PDE and delay systems. Backstepping also yields the first systematic method for control of large classes of nonlinear PDEs and for nonlinear systems with long delays.
Topics: Lyapunov stability for PDEs; boundary control of parabolic (reaction-advection-diffusion) PDEs; observers with boundary sensing; wave and beam PDEs; first-order hyperbolic (transport-dominated) PDEs; systems with input delay and predictor feedback; delayrobustness of predictor feedback; time-varying input delay; delay-adaptive predictor feedback; stabilization of nonlinear systems with long input delays; basics of motion planning for PDEs; system identification and adaptive control of PDEs; introduction to control of nonlinear PDEs.

European Embedded Control Institute

M12 - BELGRADE Verification and Correct-by-Construction Synthesis of 18/03/2013 22/03/2013 Control Protocols for Networked Systems

Richard Murray
Control and Dynamical Systems California Institute of Technology, USA

Ufuk Topcu
Control and Dynamical Systems California Institute of Technology, USA

Nok Wongpiromsarn
Singapore-MIT Alliance for Research and Technology, Singapure

Abstract of the course: Increases in fast and inexpensive computing and communications have enabled a new generation of information-rich control systems that rely on multi-threaded networked execution, distributed optimization, sensor fusion and protocol stacks in increasingly sophisticated ways. This course will provide working knowledge of a collection of methods and tools for specifying, designing and verifying control protocols for distributed systems. We combine methods from computer science (temporal logic, model checking, reactive synthesis) with those from dynamical systems and control (dynamics, stability, receding horizon control) to analyze and design partially asynchronous control protocols for continuous systems. In addition to introducing the mathematical techniques required to formulate problems and prove properties, we also describe a software toolbox, TuLiP, that is designed for analyzing and synthesizing hybrid control systems using linear temporal logic and robust performance specifications The following topics will be covered in the course: * Transition systems and automata theory * Specification of behavior using linear temporal logic * Algebraic certificates for continuous and hybrid systems * Approximation of continuous systems using discrete abstractions * Verification of (asynchronous) control protocols using model checking * Synthesis of control protocols and receding horizon temporal logic planning * Case studies in autonomous navigation and vehicle management systems

European Embedded Control Institute

M13 18/03/2013 22/03/2013

Input Saturation: Control Design and Anti-windup

Sophie Tarbouriech
LAAS-CNRS, Toulouse, France

Luca Zaccarian
LAAS-CNRS, Toulouse, France

Abstract of the course: The magnitude of the signal that an actuator can deliver is usually limited by physical or safety constraints. This limitation can be easily identified in most common devices used in the process industry, such as proportional valves, heating actuators, power amplifiers, and electromechanical actuators. Common examples of such limits are the deflection limits in aircraft actuators, the voltage limits in electrical actuators and the limits on flow volume or rate in hydraulic actuators. While such limits obviously restrict the achievable performance, if these limits are not treated carefully and if the relevant controllers do not account for them appropriately, peculiar and pernicious behaviors may be observed (aircraft crashes, Chernobyl nuclear power station meltdown).

This course addresses stability analysis and stabilization of linear systems subject to control saturation. We will discuss a first approach consists in designing a (possibly nonlinear) controller directly accounting for the saturation constraints. Then we will present the socalled anti-windup approach, where an anti-windup augmentation is inserted on an existing control system which "winds up" (performs undesirably) due to actuator saturation. The antiwindup feature is then to preserve the predesigned controller before saturation is activated and to recover stability for larger saturated responses. Anti-windup solutions differ in architecture and performance achievements. We will discuss several architectures suited for different saturation problems. Several applications will be used to illustrate the presented techniques.
Topics: Rate and magnitude saturation, standard and generalized sector conditions, stability and performance analysis with saturation, linear LMI-based controller and anti-windup designs, linear and nonlinear model recovery anti-windup design, applications

European Embedded Control Institute

M14 25/03/2013 - 29/03/2013

Traffic Modeling and Estimation at the Age of Smartphones

Alex Bayen, UC Berkeley, USA, Christian Claudel, KAUST, Saoudi Arabia, Dan Work, UIUC, USA, Sebastien Blandin, IBM Research Singapore, Aude Hofleitner, UC Berkeley, USA and Facebook Inc, http://eecs.

Abstract of the course: The recent emergence of sparsely sampled mobility data has crated new opportunity and raised challenges for control and estimation problems in intelligent urban networks. The course presents novel data filtering, modeling, estimation and control algorithms, specific to the use of smartphone data in the context of transportation and mobility. Specific implementations from the Mobile Millennium traffic information system will serve as illustrations for the course.

The following theoretical topics will be covered in the course: First order flow models: construction of the solution of the Partial Differential Equation Optimal Control theory for scalar conservation laws and Hamilton-Jacobi equations Statistical models and graphical networks: Random Markov Fields, Dynamic Bayesian Networks, Expectation Maximization algorithm Statistical inference in large scale networks: Ensemble Kalman Filter, Particle Filter Online learning of sparse models
The following applications will be covered in the course: Real-time traffic estimation on large scale highway and urban networks from crowdsourced mobile data Macroscopic behavioral traffic models on networks Modeling urban traffic on a network: a hybrid approach of queuing theory and statistical modeling

European Embedded Control Institute

M15 25/03/2013 29/03/2013

Model Predictive Control

Abstract of the course: Model Predictive Control (MPC) is the only advanced control methodology (ie more advanced than PID) which has found wide application in the process industries. It offers advantages which make it very attractive for other industries too, such as automotive and aerospace, and its use in such industries is being actively explored at present. The course will start with the basic ideas of MPC, together with some specific examples of its advantages over classical control. It will then discuss the structure of MPC controllers, present possible variations (such as non-quadratic cost functions and stabilised predictions), and deal with Jan Maciejowski important practicalities, especially disturbance Department of Engineering, feedforward and disturbance modelling. A stateUniversity of Cambridge , UK space framework will be used, but the connection with the well-known GPC framework will be made. The course will then survey the state of more advanced MPC-related research, covering efficient computation, stability and robustness, prioritisation of objectives, the use of nonlinear models, the application of MPC to hybrid systems (which contain logic or mode switches as well as continuous dynamics), and distributed MPC. The course will be illustrated throughout with examples from various applications, including flight control, spacecraft control, and paper-making.

Topics covered: 1. Basic formulation of MPC 2. Solution of MPC. The GPC formulation. 3. Other formulations of MPC. 4. Stability and tuning of MPC. 5. Robust MPC. 6. Explicit MPC. 7. Case studies & applications. 8. Recent developments & perspectives.

European Embedded Control Institute

M16 08/04/2013 12/04/2013

About Nonlinear Digital Control

Marie Dorothe Normand-Cyrot

Laboratoire des Signaux et Systmes CNRS-Univ.ParisSud-Supelec, Gif-sur-Yvette, France

Salvatore Monaco
Dipartimento di Ingegneria Informatica, Automatica e Gestionale Antonio Ruberti Sapienza Universit di Roma, Rome, Italy

Abstract of the course: To understand the effect of sampling over the control properties of a continuous-time pyisical process is preliminar to the design of a control law implemented through digital devices. Starting from this analysis equivalent and approximated sampled-data representations will be introduced. On the bases of new concepts and definitions in discrete-time, sampled-data control schemes are proposed to solve well known nonlinear control problems with reference to different classes of processes. Some case studies illustrate the computational aspects and the performances of the sampled-data control systems. Topics include: Nonlinear sampling and the properties of the-sampled data model Feedback linearization and tracking Passivity based control Lyapunov design and back-stepping techniques Delayed systems Some case studies

European Embedded Control Institute

M17 22/04/2013 26/04/2013

Event-triggered and Self-triggered Control

Maurice Heemels

Karl H. Johansson

Paulo Tabuada
Cyber-Physical Systems Laboratory Department of Electrical Engineering University of California, Los Angeles USA

ACCESS Linnaeus Centre Hybrid and Networked Systems group School of Electrical Engineering Department of Mechanical Engineering Technische Universiteit Eindhoven (TU/e) KTH Royal Institute of Technology Sweden Netherlands

Abstract of the course: Classical sampled-data control is based on periodic sensing and actuation. Due to recent developments in computer and communication technologies, a new type of resourceconstrained wireless embedded control systems is emerging. It is desirable in these systems to limit the sensor and control communication to instances when the system needs attention. This requirement calls for a paradigm shift in digital control implementations towards event-triggered and self-triggered control systems. Event-triggered control is reactive and generates sensor sampling and control actuation when, for instance, the plant state deviates more than a certain threshold from a desired value. Self-triggered control, on the other hand, is proactive and computes the next sampling or actuation instance ahead of time. As in both schemes the sampling period is varying, the vast literature on sampled-data control is no longer applicable to guarantee desirable closed-loop stability and performance properties. As a consequence, a new system theory for event-triggered and self-triggered control is needed. This course will provide an introduction to event-triggered and self-triggered control systems. Topics: The basics of event-triggered and self-triggered control will be presented showing the status and open problems in the emerging system theory for these new digital control strategies. Different design perspectives will be provided for both state feedback and output feedback eventtriggered control and various types of event-triggering mechanisms. Also distributed variants, which are suitable for large-scale control applications, will be discussed in detail. The implementation of event- and self-triggered control using existing wireless communication technology and interesting applications to wireless control in the process industry will also be presented.

European Embedded Control Institute

M18 - ISTANBUL 22/04/2013 26/04/2013

Stochastic Control with Contemporary Methods and Applications

Abstract of the course: In many applications, stochastic models are being turned to as the most effective description of control problems. This is especially true in the study of highly autonomous systems, where learning may be involved, and also in financial engineering when stochastic models have long been seen as essential. Often the combination of Markov models and ordinary differential equations provide natural and effective descriptions. However, teaching stochastic processes to students whose primary interests are in applications has long been a problem. On one hand, the subject can quickly become highly technical and if mathematical concerns are allowed to dominate there may be no time available for exploring the many interesting areas of applications. On the other hand, the treatment of stochastic calculus in a cavalier fashion leaves the student with a feeling of great uncertainty when it comes to exploring new material. This problem has become more acute as the power of the differential equation point of view has become more widely appreciated.

Roger W. Brockett
Harvard School of Engineering and Applied Sciences, USA

In this course we will resolve this dilemma with the needs of those interested in building models and designing algorithms for learning, estimation and control in mind. The approach is to start with Poisson counters and to identify the Wiener process with a certain limiting form. The Poisson counter and differential equations whose right-hand sides include the differential of Poisson counters are developed first. This leads to the construction of a sample path (Ito) representations of a continuous time jump process using Poisson counters. This point of view leads to an efficient problem solving technique and permits a unified treatment of time varying and nonlinear problems. More importantly, it provides sound intuition for stochastic differential equations and their uses without allowing the technicalities to dominate. A variety of models will be developed. For example, the wide spread interest in problems arising in speech recognition and computer vision has influenced the choice of topics in several places. Examples will be drawn from applied work in communications (wireless), artificial intelligence (path planning), physics (NMR), and other branches of mathematics.

European Embedded Control Institute

M19 - ISTANBUL 29/04/2013 03/05/2013

Symbolic Control Design of Cyber-Physical Systems

Maria Domenica Di Benedetto

Dipartimento di Ingegneria e Scienze dell'informazione e Matematica Center of Excellence DEWS University of LAquila, Italy

Giordano Pola

Alessandro Borri

Istituto di Analisi dei Sistemi ed Dipartimento di Ingegneria e Informatica "A. Ruberti" (IASI) Scienze dell'informazione e Matematica Consiglio Nazionale delle Ricerche (CNR) Center of Excellence DEWS Rome, Italy University of LAquila, Italy

Abstract of the course:

Cyber-Physical Systems (CPS) are large-scale, complex, heterogeneous, distributed and networked systems where physical processes interact with distributed computing units through communication networks. Formal approaches to the control design of these systems are relatively unexplored today. This course will present an approach to the control design of CPS based on symbolic models. Symbolic models are finite state automata where each state corresponds to an aggregate of possibly infinite continuous states and each label on the transitions to an aggregate of possibly infinite continuous inputs. We will show how the use of symbolic models provides a systematic approach to deal with control problems where software and hardware interact with the physical world through non-ideal communication networks. Efficient on-the-fly algorithms for symbolic control design will also be discussed. We will illustrate the proposed methodology on case studies.
The following topics will be covered in the course: * Transition systems, equivalence and compositionality * Approximation metrics for discrete and continuous systems * Incremental stability notions for nonlinear systems * Symbolic models for nonlinear and networked control systems * Symbolic control design * Efficient on-the-fly algorithms and case studies

European Embedded Control Institute

M20 13/05/2013 17/05/2013

Nonlinear and Adaptive Control

Romeo Ortega
Laboratoire des Signaux et Systmes CNRS-Univ.ParisSud-Supelec, Gif-sur-Yvette, France

Alessandro Astolfi
Department of Electrical and Electronic Engineering Imperial College, London, UK

Abstract of the course: Goal of this course is to present a class of recently developed control tools for the robust stabilization, by state and output feedback, of classes of nonlinear systems. These new tools enable to give an alternative formulation and solution to the stabilization problem for general nonlinear systems by means of the notions of systems immersion and manifold invariance (I&I). I&I methods are particularly suited to robustify, with respect to unmodelled dynamics, a given controller scheme. They have also proved useful in adaptive control problems, where a stabilizing controller parameterized in terms of some unknown constant vector is assumed to be known. Adaptive control applications will be the main focus of this workshop. The proposed I&I approach, which is partly reminiscent of early contributions in the area of PI adaptation, is shown to yield superior performance, when compared with classical methods, and to provide improved design flexibility and additional tuning parameters. Moreover, this approach does not require linear parameterization, it can naturally include sign constraints in the estimated parameters, and yields a new class of non-certainty equivalent control laws. From a Lyapunov perspective this is the first systematic method to construct non-separable Lyapunov functions, i.e. Lyapunov functions containing cross terms depending upon the system state and the parameters estimation error, without assuming a specific structure of the nonlinear system to be controlled. The theory is illustrated by means of applications and experimental results. In particular, solutions to the adaptive stabilization problem for classes of power converters and electrical machines and for the problem of visual servoing of a planar robot are discussed.

Topics include: - State feedback stabilization and adaptive control via immersion and invariance - Output feedback adaptive control via immersion and invariance - Applications in adaptive control - Applications to electromechanical systems - Open problems

European Embedded Control Institute

M21 13/05/2013 17/05/2013

Distributed Control

Abstract of the course: Over the past decade there has been growing in interest in distributed control problems of alltypes. Among these are consensus and flocking problems, the multi-agent rendezvous problem, distributed averaging and the distributed control of multi-agent formations. The aim of these lectures is to explain what these problems are and to discuss their solutions. Related concepts from spectral graph theory, rigid graph theory, nonhomogeneous Markov chain theory, stability theory,and linear system theory will be covered.

A. Stephen Morse
Department of Electrical Engineering Yale University, USA

Topics include: 1. Flocking and consensus 2. Distributed averaging via broadcasting 3. Gossiping and double linear iterations 4. Multi-agent rendezvous 5. Control of formations 6. Contraction coefficients 7. Convergence rates 8. Asynchronous behavior 9. Stochastic matrices, graph composition, rigid graphs

European Embedded Control Institute

M22 20/05/2013 24/05/2013

Extremum Seeking Control: Analysis and Design

Abstract of the course: A great majority of control engineering design methods deals with the analysis and design of transient behaviour in closed-loop systems. However, for many engineered systems, a crucial aspect of their operation is that their steady-state behaviour is best in some sense. Extremum seeking techniques provide a systematic methodology for optimization of the steady-state behaviour via closed-loop techniques in cases when the model of the plant and/or the cost to optimize are not known to the designer. This on-line optimization methodology has been successfully used in a range of engineering applications but only recently we have developed appropriate techniques and tools to systematically design and analyze large classes of such systems. This subject presents state-of-the-art methods and techniques for extremum seeking control. We will make direct connections to off-line continuous and discrete nonlinear programming, adaptive control and present detailed stability analysis, as well as controller tuning guidelines that are invaluable to practicing engineers.

Dragan Nesic
Department of Electrical and Electronic Eng. The University of Melbourne, Australia

Singular perturbations Averaging Lyapunov stability of continous-time and discrete-time nonlinear systems Continous-time and discrete-time off-line optimization (nonlinear programming) with examples (e.g. gradient methods, Newton schemes, etc) Continuous-time extremum seeking (black box and gray box approaches) Convergence analysis and tuning guidelines of continuous schemes Discrete-time extremum seeking Convergence analysis and tuning guidelines of discrete-time schemes

European Embedded Control Institute

M23 20/05/2013 24/05/2013

Robust Hybrid Control Systems

Abstract of the course: Hybrid control systems arise when controlling nonlinear systems with hybrid control algorithms algorithms that involve logic variables, timers, computer program, and in general, states experiencing jumps at certain events and also when controlling systems that are themselves hybrid. Recent technological advances allowing for and utilizing the interplay between digital systems with the analog world (e.g., embedded computers, sensor networks, etc.) have increased the demand for a theory applicable to the resulting systems, which are of hybrid nature, and for design techniques that may guarantee, through hybrid control, performance, safety, and recovery specifications even in the presence of uncertainty. In the workshop, we will present recent advances in the theory and design of hybrid control systems, with focus on robustness properties.

Ricardo G. Sanfelice
Dept. Aerospace & Mechanical Engineering University of Arizona, USA

Topics: In this course, we will present a general modeling framework for hybrid systems and relevant modern mathematical tools. Next, we will introduce asymptotic stability and its robustness, and describe systematic tools like Lyapunov functions and invariance principles. The power of hybrid control for (robust) stabilization of general nonlinear systems will be displayed in applications including control of robotic manipulators, autonomous vehicles, and juggling systems

European Embedded Control Institute

M24 LAQUILA 20/05/2013 24/05/2013

Optimality, Stabilization, and Feedback in Nonlinear Control

Abstract of the course:

This course presents some modern tools for treating truly nonlinear control problems, including non smooth calculus and discontinuous feedback. The need for such tools will be motivated, and applications will be made to central issues in optimal and stabilizing control. The context throughout is that of systems of ordinary differential equations, and the level will be that of a graduate course intended for a general control audience.

Francis Clarke
Institut Camille Jordan Universit Claude Bernard Lyon 1, France

Topics include: 1. 2. 3. 4. 5. Dynamic optimization: from the calculus of variations to the Pontryagin Maximum Principle Some constructs of nonsmooth analysis, and why we need them Lyapunov functions, classical to modern Discontinuous feedback for stabilization Sliding modes and hybrid systems

European Embedded Control Institute

M25- LAQUILA 27/05/2013 31/05/2013

Modeling and Estimation for control

Abstract of the course: The objective of this class is to introduce multiphysics models for complex dynamical systems, with different modeling, identification and estimation methods. The purpose of such models is to include physical knowledge of the systems as well as experimental data, and to allow for preliminary system design, predictive diagnostic and real-time control.

Emmanuel Witrant
Dpartement Automatique CNRS Gipsa-Lab, Grenoble, France

Topics :

1. Introduction to modeling Physical modeling 2. Principles of physical modeling 3. Some Basic Relationships in Physics. 4. Bond Graphs Simulation 5. Computer-Aided Modeling 6. Modeling and Simulation in Scilab System identification 7. Experiment Design for System Identification 8. Non-parametric Identification 9. Parameter Estimation in Linear Models 10. System Identification Principles and Model Validation 11. Nonlinear Black-box Identification Towards process supervision 12. Recursive Estimation Methods
For more details, see

European Embedded Control Institute

M26 27/05/2013 31/05/2013

Switched Systems and Control

Abstract of the course:

Switched systems are dynamical systems described by a family of continuous-time systems and a rule that orchestrates the switching between them. Such systems are interesting objects for theoretical study and provide realistic models suitable for many applications. This course will examine switched systems from a control-theoretic perspective. The main focus will be on stability analysis and control synthesis of systems that combine continuous dynamics with switching events. In the analysis part of the course, we will develop stability theory for switched systems; properties beyond traditional stability, such as invertibility and input-to-state stability, will also be discussed. In the synthesis part, we will investigate several important classes of control problems for which the logic-based switching paradigm emerges as a natural solution.

Daniel Liberzon
Coordinated Science Laboratory University of Illinois at Urbana-Champaign, USA

Topics include: Single and multiple Lyapunov functions

Stability criteria based on commutation relations

Stability under slow switching Switched systems with inputs and outputs Control of nonholonomic systems

Quantized feedback control

Switching adaptive control