2010 IEEE Multi-Conference
on Systems and Control

September 8-10, 2010
Yokohama, a port city on Tokyo Bay

Tutorial Sessions

CCA      WeA05, Wednesday, September 8, 10:30-11:50, Room 413

Mathematical Modeling in Systems Biology

Organizers: Michael Ederer and Oliver Sawodny (Univ. of Stuttgart)

Abstract: Systems Biology is nowadays a well established and still rapidly growing research field. It emerged from the application of systems theory on biological systems. This cooperation gives valuable input as well as new challenges to both sides.
    Mathematical modeling and model analysis of biochemical reaction networks gets increasingly important for biological research and its application in medicine and bioprocess engineering. In the first tutorial, we will give an overview of different modeling techniques. Further, we will discuss the special role of mathematical modeling in biology. Whereas in engineering, models are mainly means to reach a design task (e.g. controller design), models in biology are often a value by itself. Models allow coherently and comprehensively integrating and developing our knowledge of a biological system.
    The following tutorials will be focusing on parameter identification and bifurcations in the dynamics of biochemical networks. Parameter identification for biochemical reaction networks is a challenge and needs special approaches as the relation of available measurement data to unknown variables is often demanding. Complex behavior in the dynamics of biochemical reaction networks is commonly governed by bifurcations of equilibria and limit cycles. In this tutorial, we discuss two frequent types of bifurcations, the saddle-node and the Hopf bifurcation, and their relation to bistability and oscillations in biochemical networks. We also present examples from molecular biology where the dynamical behavior of specific networks is defined by these bifurcations.
  • Mathematical Modeling in Systems Biology (20 minutes)
      Michael Ederer
  • Advances in Global Parameter Estimation Approaches for Biochemical Reaction Networks: An Overview (40 minutes)
      Philipp Rumschinski, Steffen Borchers, Anton Savchenko, and Rolf Findeisen (OVGU Univ. Magdeburg)
  • Generic Bifurcations in the Dynamics of Biochemical Networks (20 minutes)
      Nicole Radde (Univ. of Stuttgart)

CACSD      ThA05, Thursday, September 9, 10:30-12:30, Room 413

The Role of Systems Theory in Computer Vision

Organizer: Mario Sznaier (Northeastern Univ.)

Abstract: Computer vision is uniquely positioned to enhance the quality of life for large segments of the general public. Aware sensors endowed with tracking and scene analysis capabilities can prevent crime, reduce time response to emergency scenes and allow elderly people to continue living independently. Enhanced imaging methods can substantially reduce the amount of radiation required in medical imaging procedures and in cancer therapy. Moreover, the investment required to accomplish these goals is relatively modest, since the number of imaging sensors already deployed and networked is large (for instance over 10,000 in Manhattan alone) and is exponentially increasing since the introduction of camera cell phones capable of broadcasting and sharing live video feeds in real time. The challenge now is to develop a theoretical framework that allows for robustly processing this vast amount of information, within the constraints imposed by the need for real time operation in dynamic, partially stochastic scenarios.
    On the other hand, during the past decades systems theory has achieved a high degree of maturity, leading to powerful and sophisticated tools that have allowed for solving difficult practical problems. Central to the success of this effort is a viewpoint that emphasizes both robustness and complexity issues, seeking for computationally tractable solutions, or in cases where the underlying problem is intrinsically hard, for tractable relaxation with suboptimality certificates. The goal of this session is to illustrate the central role that these ideas can play in developing a comprehensive robust dynamic vision framework. The talks show how the use of system theoretic tools in computer vision have led to either new theoretical results, solutions to open problems or a better understanding of the phenomena involved, while offering a good cross-section of the current state of the field and pointing to problems that remain open. 1
  • Passivity-based Visual Motion Observer: From Theory to Distributed Algorithms (40 minutes)
      Takeshi Hatanaka and Masayuki Fujita (Tokyo Inst. of Tech.)
  • Estimation Theory and Tracking of Deformable Objects (40 minutes)
      Patricio A. Vela and Ibrahima J. Ndiour (Georgia Inst. of Tech.)
  • Dynamics Based Extraction of Information Sparsely Encoded in High Dimensional Data Streams (40 minutes)
      Mario Sznaier and Octavia Camps (Northeastern Univ.)

CCA      ThB08, Thursday, September 9, 14:00-16:00, Room 421

Environmental and Energy Control Systems

Organizers: Toru Namerikawa (Keio Univ.), Eugenio Schuster (Lehigh Univ.), and Tomohisa Hayakawa (Tokyo Inst. of Tech.)

    Abstract: Due to the large demands on the innovation of technology for renewable energy and the increasing interests in developing a variety of new frameworks of keeping the world environment healthy, research on controling energy systems has been attracting more and more attention. To facilitate this trend in the control community, we organize the tutorial session on the "Environmental and Energy Control Systems" at 2010 MSC. Specifically, the first two speakers talk about control methods in power generation and the next two speakers deal with fundamental framework of controling power grids. Finally, last two speakers explain recent technology in Toyota plug-in hybrid cars and control operation of heating/cooling systems of buildings.

  • Power Generation and Control (20 minutes)
      Eugenio Schuster
  • Output Optimization of Wind Energy Generation Systems via Extremum Seeking Control (20 minutes)
      Tomohisa Hayakawa
  • Renewable Energy Grid Integration (20 minutes)
      Rush D. Robinett III and David Wilson (Sandia National Lab.)
  • Electric Demand Prediction and Predictive Control of Micro Grid (20 minutes)
      Toru Namerikawa
  • The Newly Developed Toyota Plug-in Hybrid System (20 minutes)
      Toshifumi Takaoka (Toyota Motor Corp.)
  • Energy-efficient Control of Building Heating/Cooling Systems (20 minutes)
      Alessandro Beghi (Univ. di Padova)

ISIC      FrA05, Friday, September 10, 10:30-12:30, Room 413

Implicit Learning Control Through a RISE Architecture: Theory and Applications

Organizers: Parag Patre (NASA Langley Res. Center) and Warren Dixon (Univ. of Florida)

    Abstract: A dynamic robust control strategy (coined RISE - Robust Integral of the Signum of the Error) was recently developed as the first class of continuous controllers that can yield asymptotic tracking for a general class of uncertain nonlinear systems with sufficiently smooth bounded disturbances. This control method is described as a type of implicit learning controller since the learning capability of the controller is implicit and does not require an explicit adaptive or learning law. A robust control term is used to exactly learn/identify uncertain dynamics in the system through a nonlinear differential equation.
        This tutorial session provides an introduction to the RISE control method by illustrating the implicit learning capacity in comparison with other existing methods. Adaptive controllers and neural networks are typical alternative intelligent controllers that can compensate for uncertain dynamics. However, typical adaptive methods can only compensate for linear in the constant parameters uncertainty, and neural networks have an inherent function reconstruction error that yield an ultimately bounded approximate tracking result. The second section of the tutorial session illustrates how the RISE control method can be fused with these traditional intelligent control methods to yield improved asymptotic performance. The third section of the tutorial examines the ability of the RISE controller to asymptotically yield an optimal controller. Specifically, the RISE method is used as an adjuvant to an optimal controller to produce a solution to a Hamilton Jacobi Bellman equation for an uncertain nonlinear system. The session concludes with an overview of various applications of the RISE controller to systems such as: hypersonic aircraft, electrical stimulation of human skeletal muscle, friction identification, robotics, and imaged based estimation.

  • Introduction to RISE and Implicit Learning (40 minutes)
      Parag Patre and Warren Dixon
  • Adaptive and Neural Network-Based Control (20 minutes)
      Parag Patre and Warren Dixon
  • Direct Optimal Control (20 minutes)
      Warren Dixon and Parag Patre
  • RISE Applications (40 minutes)
      Warren Dixon and Parag Patre

CACSD      FrB01, Friday, September 10, 14:00-16:00, Room 417

Recent Advances in Model Reduction of Large-Scale Systems

Organizer: Athanasios C. Antoulas (Rice Univ.)

Abstract: Model order reduction seeks to replace a large- scale system by a system of substantially lower dimension that has nearly the same response characteristics. The methods used to achieve this goal can be classified in two broad categories, namely, SVD- (Singular Value Decomposition) based methods and Krylov-based or moment matching methods. In this session we will provide an overview of recent advances in reduction methods within these two broad categories.
  • Balanced Truncation Model Reduction (40 minutes)
      Timo Reis (Tech. Univ. Hamburg-Harburg)
  • Interpolatory Model Reduction (40 minutes)
      Serkan Gugercin (Virginia Tech.)
  • Model Reduction From Data (40 minutes)
      Athanasios C. Antoulas