|
Workshop Lineup
S-2 is moved to Monday afternoon at the organizer's request.
There will be no registration fee charged for M-2 but people should still register in order to get an accurate count.
Workshops will take place from 8:00 AM to 5:00 PM on Sunday, December 7th and Monday, December 8th, 2003 at the conference hotel. Workshop T-1 is a 2-day workshop. Workshops S-1, S-2, and M-1 are 1/2-day workshops. All others are 1-day workshops. Registration fees are as follows:
|
|
Attending CDC
|
Not attending CDC
|
Student/Retiree
|
|
1/2-day
|
$100
|
$175
|
$50
|
|
1-day
|
$150
|
$275
|
$70
|
|
2-day
|
$200
|
$350
|
$100
|
BACK TO TOP
Workshop S-1 (1/2 Day): Feedback Control of Biped Walking Robots
Sunday, 8:00 AM - 12:00 Noon
Organizers:
J. W. Grizzle, University of Michigan
C. Canudas de Wit, Laboratoire d'Automatique de Grenoble
Additional Participants:
E.R. Westervelt, The Ohio State University
M.W. Spong, University of Illinois at Urbana-Champaign
Summary
Over the past several years, a group of people in the control community have begun to investigate the systematic design of feedback controllers for bipedal robots, with emphasis on careful model formulation and analysis of the stability properties of the closed-loop system. This workshop brings together four of these researchers to present their work in a tutorial fashion to the control community.
The first three lectures will be centered around a particular biped that was specifically designed to advance the fundamental understanding of controlled legged locomotion. In particular, the mechanism was designed to allow for high speed walking and running. The robot's lateral stabilization is assured by a rotating bar, and thus only 2-D motion in the sagittal plane is considered. Except for this, the prototype captures the main difficulties inherent in this type of nonlinear system: underactuation (no feet), variable structure (the state dimension varies as a function of the motion phase), and state jumps (sudden state variations resulting from impacts with the ground). For more details, see http://www.eecs.umich.edu/~grizzle/papers/robotics.html or
http://www-lag.ensieg.inpg.fr/PRC-Bipedes/English/index.php.
The last lecture will present a detailed discussion of the notion of passive walking and its relation to passivity-based and hybrid nonlinear control. Passive walking has been a well-known and intriguing subject in bipedal locomotion ever since the pioneering work of McGeer, who showed more than a decade ago that it is possible to construct mechanisms that exhibit locomotion down shallow inclines without sensing or actuation of any kind. Such mechanisms typically exhibit extreme sensitivity to the ground slope as well as to initial conditions and external disturbances. In this lecture, the notions of symmetry in Lagrangian systems will be introduced and used to show how potential energy shaping can be used to completely remove the sensitivity to ground slope. In addition, total energy shaping and switching control will be used to mitigate the sensitivity to initial conditions and external disturbances. Simulation and animation results will be shown for a planar biped with torso and knees.
BACK TO TOP
Workshop S-2 (1/2 Day): Power Control in Wireless Communication Networks
Monday, 1:00 PM - 5:00 PM
Organizers:
Zoran Gajic, Rutgers University
Sarah Koskie, Indiana University-Purdue University, Indianapolis
Summary
Signal power is one of the most important commodities in signal transmission over communication channels. During the last ten years, the problem of efficient mobile power control in wireless communication networks has become a central research problem that has resulted in several hundred journal and conference papers. It has recently attracted broad interest from experienced control engineering researchers (for example, Basar, Bertsekas, Caines, Ephremides, Gajic, Hollot, Srikant). For high quality of signal transmission a high signal power to interference power ratio is required; but in wireless channels, increasing the signal power of one mobile increases interference to other mobiles using the same communication channel (cochannel users). Even more, high mobile powers drain mobile batteries quickly and require frequent battery recharging. Balancing mobile powers so that each cochannel user has a satisfactory quality of service is a task that must be addressed by the base station (the central decision maker) though it is desirable that the balancing be achieved in a distributed manner using only local information for each mobile. Since mobile powers are updated 800 time per second (in a currently implemented IS-95 protocol), the problem can be described by a set of difference equations, which define the dynamics of mobile power adjustments (distribution). The search for efficient mobile power distribution schemes leads to several interesting estimation- and optimization-type control problems, which, in general, require resolving either static or dynamic conflict game situations. Those problems have been studied recently in deterministic and stochastic setups using both classical and modern control theory techniques.
Part I (2 hours): A survey of early results on power control in wireless communications that will introduce problem formulations, identify research challenges, and indicate some early solutions, to be presented by Professor Zoran Gajic.
This part will start with the power control problem formulation that originated within satellite communications and evolved into the power control problem formulation for wireless networks at the beginning of the nineties. The so-called power balancing solution will be defined and its relation to the signal-to-interference-ratio (SIR) will be fully explained. Several SIR-based power control algorithms that have a foundation in numerical linear algebra will be discussed. Other results on power control for wireless networks (not necessarily SIR-based) obtained during the nineties will be summarized with emphasis on the control engineering solutions, including a simple and elegant classic PI controller solution. The presentation in this part will be done in a chronological order.
Part II (2 hours): A survey of recent results applying control theory to power control in wireless communications, to be presented by Assistant Professor Sarah Koskie, a recent graduate student of Professor Gajic.
These techniques will include 1) optimal control theory solutions for different performance costs, 2) static and dynamic Nash game problem formulations and solutions, and 3) estimator-based techniques for use in a stochastic problem formulation. These last will include the Kalman and H-infinity filters, which may be used to estimate either interference signals or channel variations. This presentation will form a coherent series of mini lectures presenting alternative problem formulations and solutions for SIR-based power control problems in wireless networks.
BACK TO TOP
Workshop S-5: Nonlinear Predictive Control
Sunday, 8:00 AM - 5:00 PM
Organizers:
Mike Grimble and Andrzej Ordys, Univ. of Strathclyde
Additional Participants:
Dr. Brent Brunell, General Electric, USA
Professor Jan Maciejowski, University of Cambridge, Cambridge, UK
Professor Basil Kouvaritakis, University of Oxford, Oxford, UK
Dr. B. de Schutter, Delf University of Technology, Belgium
Professor Joseph Bentsman, University of Illinois, USA
Professor Dale Seborg, University of Santa Barbara, Santa Barbara, USA
Summary
The aim of this workshop is to provide a tutorial introduction to new techniques in non-linear predictive control and to take towards future developments in the subject. The workshop will motivate the need for research in this area and provide a basic introduction to some of more successful methods being developed. This will be emphasized by a broad discussion of different fields of application of nonlinear predictive control.
In recent years, an increased interest in non-linear control systems can be observed. This is driven by two factors. From one side, computational power available nowadays is much larger than it was before, making complicated computations more feasible in real time. From the other side, there is a need for optimization of profits (and accuracy) in competitive industrial environment and this means utilization of the full (non-linear) characteristics of available processes.
One of the most promising techniques, especially in (but not limited to) process industry, is non-linear predictive control.
The characteristic feature of this Workshop is a large number of invited speakers, which should provide a very wide and unbiased overview of the current state of the art in the subject of non-linear predictive control. The Workshop will start with an introductory talk describing the motivation - the industrial need for non-linear predictive techniques.
In subsequent presentations, the main theoretical issues of non-linear predictive control will be addressed. This will include stability and robustness, computational feasibility of the algorithms and prediction methods.
The industrial application examples will follow, illustrating the benefits of nonlinear predictive control.
The Workshop will conclude with the presentation on current and future trends in research on non-linear predictive control, which will be followed by a round table discussion.
BACK TO TOP
Workshop M-1 (1/2 Day): Open Issues in Analysis and Control of Power and Energy Processing Systems
Monday, 8:00 AM - 12:00 Noon
Organizers:
Ian Hiskens, Univ. of Wisconsin
Alex Stankovich, Northeastern Univ.
Additional Participants:
Eyad Abed, University of Maryland
Iven Mareels, University of Melbourne
Romeo Ortega, Supelec
Seth Sanders, UC-Berkeley
Jeff Lang, MIT
Ken Loparo, Case Western Reserve University
Summary
Evolving technology in the generation, transmission, and utilization of electric energy is driving a reassessment of traditional analysis and control techniques. Higher penetration of distributed generation places greater emphasis on the dynamic response of lower voltage systems. Coordination of large numbers of small generators, in the presence if significant load uncertainty, is a challenging distributed control problem. Flexible AC transmission systems (FACTS) devices offer control-based improvements in transmission system utilization. However, advances in control design methods are required. Power electronic converters are driving a revolution in electrical energy utilization, as well as underlying distributed generation. These energy conversion systems exhibit hybrid (continuous/discrete) dynamics. At the utilization extreme, the physical properties of MEMS devices ensure nontrivial control design issues. The aim of this workshop is to bring together leading researchers in these areas to provide a forum for the wider systems community to encounter interesting, open research questions in these important application areas.
BACK TO TOP
Workshop M-2: Cross-Disciplinary Research and the Role of Industry (NSF Sponsored)
Monday, 8:00 AM - 5:00 PM
Organizers:
Bob Barmish, University of Wisconsin
John Baras, University of Maryland
Lennart Ljung, Linkoping University
Richard Murray, Caltech
Mike Grimble, University of Strathclyde
Additional Participants:
Pramod Khargonekar, University of Florida
Kenwood Hall, Rockwell Automation
Kishan Baheti, NSF
John Kenyan, Hughes Network Systems
Summary
The purpose of this workshop is to explore mechanisms for cross-disciplinary research in the control community, particularly the interaction with industry. Modern control involves the development and implementation of a wide variety of very complex engineering systems and the control community has been a major source of training for people who embrace a systems perspective. All modern systems design is done in interdisciplinary teams and it requires certain skills to understand how to effectively interact with domain experts from a wide variety of disciplines. Industry plays a key role in the community in this regard, providing both the success stories of how to integrate interdisciplinary teams and opportunities for students and faculty in universities to interact with such teams.
There is no registration fee for this workshop, but participants must preregister through the conference registration form (so that we can get an accurate count of the attendees).
A limited amount of funding is available to support one day's lodging for participants in the workshop. Additional information will be available here on 1 Sep 2003.
BACK TO TOP
Workshop M-3: Semidefinite Programming Relaxations and Algebraic Optimization in Control
Monday, 8:00 AM - 5:00 PM
Organizers:
Pablo Parrilo, ETH Zščrich
Sanjay Lall, Stanford University
Summary
This workshop focuses on computation of certified guarantees for complex control systems. Such guarantees have been a unifying theme in much control systems research in the past decade, and the applications of this notion are widespread in the literature. We intend to focus on the use of convex optimization and algebraic duality, the union of which has undergone recent rapid development and which together have an extremely large set of concrete applications.
The core of the workshop will be original material presented during the morning session. For this we will include both standard notions as well as recent developments in convexity and duality, with an emphasis on complexity and computational approaches. The workshop development will be punctuated with a series of detailed numerical examples to illustrate the approach used. The afternoon session will be a more detailed development of significant concrete applications illustrating the new ideas and concepts. The main topical areas are:
Basic notions of convexity and convex optimization. Standard classes of problems, including linear, second-order cone, and semidefinite programs. Duality as proof certificates
Computational Complexity. Models of Computation, the Turing and the real model. NP vs. co-NP problems and their relationship to duality. NP-completeness and NP-hardness, and the polynomial hierarchy.
Algebraic and symbolic computation. The difference between symbolic and numerical approaches. Linking convexity-based and algebraic methods.
Polynomial inequalities and semialgebraic sets. Polynomial nonnegativity and sums of squares. Algebraic notions of duality
The Positivstellensatz, polynomial certificates, and duality. Special representations. SDP relaxations for semialgebraic problems and combinatorial problems. Randomization schemes and guaranteed solutions.
Lyapunov functions and certificates. Interpretation of dual solutions. Effective computation and structure exploitation.
Applications: combinatorial optimization, robust optimization, stochastic models, model invalidation, biology-oriented applications, hybrid systems, quantum mechanics, decentralized control, geometric optimization.
BACK TO TOP
|
|
|
