Sunday, December 7, 2008 in Cancun Mexico
Schedule
8:15-8:30 Welcome and Opening Remarks
8:30-10:30 Presentations
11:00-12:00 Presentations
2:00-4:00 Presentations
4:30-5:30 Presentations
7:30 Dinner
Abstracts, Papers and Presentations (alphabetically by last name of speaker)
Multivariable Zero-free Transfer Functions, and their Application in Economic Modeling
Brian Anderson and Manfred Deistler
Abstract, Presentation
A Study of Feedback Fundamentals
Panos Antsaklis
Abstract, Presentation
Some Notes on Realization Theory
Shankar Bhattacharyya
Abstract, Presentation
New Trends in Neuro-Fuzzy Adaptive Control Schemes
Manolis A. Christodoulou and Yiannis S. Boutalis
Abstract, Paper, Presentation
Zeros and Zero Dynamics for Linear, Time-delay Systems
G. Conte and A. M. Perdon
Abstract, Paper, Presentation
Recent Advances in Positive Systems: Evolution to the Servo Problem
Bartek Roszak and Edward Davison
Abstract, Presentation
Sensor Classification for Control and Diagnosis Problems, a Structural Approach
C. Commault, J.M. Dion and D.H. Trinh
Abstract, Paper, Presentation
Control in Semiconductor Wafer Manufacturing
Abbas Emami-Naeini and Dick de Roover
Abstract, Paper
A Simple Polynomial Approach to Nonlinear Control
Mike Grimble and Pawel Majecki
Abstract, Presentation
Medical Robots and their Control Paradigms
Peter Kazanzides
Abstract, Presentation
Deadbeat Response is l2 Optimal
Vladimir Kucera
Abstract, Paper, Presentation
Neural Network-Based Adaptive Optimal Controller - A Continuous-Time Formulation
D. Vrabie, F. Lewis and D. Levine
Abstract, Paper, Presentation
On Strongly Stabilizing Controller Synthesis
Hitay Özbay
Abstract, Presentation
From Centralized to Distributed Fault Detection: an Adaptive Approximation Approach
Thomas Parisini and Marios M. Polycarpou
Abstract, Presentation
Rotational Motion with Almost Global Stability
Jose Vasconcelos, Anders Rantzer, Carlos Silvestre, Paulo Oliveira
Abstract, Presentation
Numerical Algorithms for Polynomial Plus/Minus Factorization
Martin Hromcik and Michael Sebek
Abstract, Paper, Presentation
Some Results on Input Classes for Identifiability
Eduardo D. Sontag
Abstract, Presentation
Micromanipulation Using Vision and Force Feedback
Hakan Bilen and Mustafa Unel
Abstract, Presentation
Abstracts and Contact Information for the above Speakers and Titles
Multivariable Zero-free Transfer Functions, and their Application in Economic Modeling
Brian D O Anderson, Australian National University
Manfred Deistler, Technical University of Vienna
brian.anderson@anu.edu.au, deistler@tuwien.ac.at
http://users.rsise.anu.edu.au/~briandoa/
http://www.eos.tuwien.ac.at/Oeko/MDeistler/
Central banks and funds investment managers work with mathematical models. In recent years, a new class of model has come into prominence—generalized dynamic factor models. These are characterized by having a modest number of inputs, corresponding to key economic variables and industry-sector-wide variables for central banks and funds managers respectively, and a large number of outputs, economic time series data or individual stock price movements for example. It is common to postulate that the input variables are linked to the output variables by a finite-dimensional linear discrete-time dynamic model, the outputs of which are corrupted by noise to yield the measured data. The key problems faced by central banks or funds managers are model fitting given the output data (but not the input data), and using the model for prediction purposes.
These are essentially tasks usually considered by those practicing identification and time series modelling. Nevertheless there is considerable underlying linear system theory. This flows from the fact that the underlying transfer function matrix is tall.
This talk will describe a number of consequences of this seemingly trivial fact. For example, a tall transfer function of known McMillan degree but otherwise generic has no zeros, finite or infinite. A finite sequence of output data in the discrete time case allows recovery of a finite sequence of input data, without knowledge of the initial state. Canonical state-variable forms take on a special format.
Key references outlining this work are as follows: Anderson, B.D.O., Deistler, M., Generalized linear dynamic factor models—a structure theory, CDC 2008, to appear Anderson, B.D.O. and Deistler, M., Properties of zero-free transfer function matrices. SICE Journal of Control, Measurement and System Integration, to appear
back to top
A Study of Feedback Fundamentals
Panos Antsaklis
Department of Electrical Engineering, University of Notre Dame
antsaklis.1@nd.edu
http://www.nd.edu/~pantsakl
In the area of Systems and Control theory, the emphasis has been on designing feedback controllers given a model of the process to be controlled and a number of control specifications. Many powerful methodologies have been introduced in the past half century to design controllers that stabilize and achieve desired performance in a robust way, being tolerant to certain class of plant parameter variations and external disturbances. Feedback or closed loop control, instead of feed-forward or open loop control, is used because of uncertainties in the plant and its environment. Significantly less effort has been spent in the past half century on understanding exactly how and why feedback works so well not only in the control of engineered systems but in natural systems as well. What are the fundamental principles, the fundamental mechanisms, which make feedback control so powerfully effective? These fundamental mechanisms should be independent of the particular type of mathematical models used, that is the system may be described by differential equations, by automata, by logic expressions, by natural language since we do know that feedback is ubiquitous and works! What are these fundamental properties that are present everywhere? This talk will raise many questions and I hope will provide a few answers as well.
back to top
Some Notes on Realization Theory
S.P.Bhattacharyya
Dept. of Electrical and Computer Engg.,
Texas A & M University
bhatt@ece.tamu.edu
http://www.ece.tamu.edu/People/bios/bbhattac.html
In this talk, prepared in honor of Prof. Wolovich’s 70th birthday, we discuss some issues related to minimal and non-minimal realizations and their role in controller synthesis. In particular, it is pointed out that realizations may be generically minimal or fragile with respect to system parameters. The latter class of systems are minimal only on an algebraic variety. Reliable control and stabilization can be achieved in such cases by first perturbing the system away from this variety to restore its locally maximal order and to base controller design based on this perturbed model. Examples illustrating this departure from the conventional wisdom of always using minimal realizations, will be included in the talk.
back to top
New Trends in Neuro-Fuzzy Adaptive Control Schemes
Manolis A. Christodoulou and Yiannis S. Boutalis*
Department of Electronic and Computer Engineering
Technical University of Crete, Greece
*Department of Electrical and Computer Engineering
University of Thraki, Greece & University of Erlangen-Nuremberg, Germany
manolis@ece.tuc.gr
ybout@ee.duth.gr
We use a new definition of Neuro-Fuzzy Dynamical Systems, using the concept of Fuzzy Dynamical Systems (FDS) in conjunction with High Order Neural Network Functions (F-HONNFs). The dynamical System is assumed nonlinear and totally unknown. We first propose its approximation by a special form of a fuzzy dynamical system (FDS) and in the sequel the fuzzy rules are approximated by appropriate HONNF’s. Thus the identification scheme leads to a Recurrent High Order Neural Network, which however, takes into account the fuzzy output partitions of the initial FDS. The proposed scheme does not require a priori experts’ information on the number and type of input variable membership functions, making it less vulnerable to initial design assumptions.
After the identification process we adaptively control the system either directly or indirectly. By doing so, we present weight updating laws for the involved HONNs. With rigorous proofs we guarantee that the errors converge to zero exponentially fast, or at least become uniformly ultimately bounded. At the same time we guarantee stability by proving that all signals in the closed loop remain bounded.
During both the identification and control process we assume, first that we know the centers and shapes of membership functions, and we identify the HONN parameters in which case we get a directional variation. Thus in order to guarantee existence of the control law, we define a new method replacing the well known projection, which is termed parameter hopping and thus we rigorously prove existence of the control law, guaranteeing stability properties.
In the sequel we also assume, that both membership function centers and HONN parameters are identifiable, ending up with a bilinear multivariable adaptive control problem which is solved for the first time in the adaptive controls literature. In this last case we are able to adaptively controlling the system, while the updating laws automatically update the parameters as well as the centers of the membership functions, placing them optimally in space. The only requirement is that we know the signs of the centers, a condition that may be relaxed in our future research work using a method similar to the Nussbaum gain for scalar-vector bilinear cases.
Simulations illustrate the potency of the method and comparisons with conventional approaches are given. The simulation tests are based on benchmark examples. Also, the applicability of the method is tested on a DC Motor system where it is shown that by following the proposed procedure one can obtain asymptotic regulation.
back to top
Zeros and Zero Dynamics for Linear, Time-delay Systems
G. Conte and A. M. Perdon
Dipartimento di Ingegneria Informatica, Gestionale e dell’Automazione
Università Politecnica delle Marche, Ancona , Italy
gconte@univpm.it, perdon@univpm.it
The notion of zero of a linear, dynamical system can be investigated and studied from many different points of view. In particular, the algebraic approach proposed some time ago by M. Sain and B. Wyman provides conceptual and practical tools for generalizing the notion of zero to several classes of dynamical systems, notably to systems with coefficients in a ring. By representing time delay systems as systems with coefficients in a ring, it is possible to introduce a suitable notion of zeros and of zero dynamics in the time delay framework, that turns out to be useful for studying interesting control problems. In this paper, we discuss such notions and their features, both from an algebraic point of view and from a geometric one, highlighting the relationship between zeros and controlled invariance subspaces of the state space. In addition, we investigate the role of zeros and zero dynamics in the design of feedback control loops and in inversion problems involving time delay systems and we show the practical use of those notions in dealing with tracking problems in the time-delay framework.
back to top
Recent Advances in Positive Systems: Evolution to the Servo Problem
Bartek Roszak and Edward J Davison*
Dept of Electrical and Computer Engineering
University of Toronto
ted@control.utoronto.ca
http://www.control.toronto.edu/people/profs/ted/ted.html
The servomechanism problem has been of importance in the emergence of LTI systems over the span of several decades. However for the class of “positive systems” , there is a gap in the research literature in that the tracking and disturbance rejection problem, although of great importance, has been ignored .Thus the goal of this talk is to give an overview of recent results obtained for the tracking and disturbance problem for positive LTI systems.
A positive LTI system is an LTI system with the imposed constraints that the state, output and/or input variables be nonnegative for all time. The motivation for studying these systems is that the nonnegative property occurs quite frequently in various applications and in nature, e.g. they immediately arise in hydrology, engineering stocking, baking ovens, furnace systems, building temperature control systems, in almost all areas of biology ….
The talk will present various results discussing the servomechanism problem under measurable or unmeasurable disturbances for both known (or unknown) positive LTI systems.
back to top
Sensor Classification for Control and Diagnosis Problems, a Structural Approach
C. Commault, J.M. Dion and D.H. Trinh
GIPSA-Lab Grenoble
Department of automatic control
Institut Polytechnique de Grenoble, France
jean-michel.dion@gipsa-lab.inpg.fr
http://www.gipsa-lab.inpg.fr/
Control and diagnosis of dynamical systems require measures or estimation of system variables via sensors. In this paper we consider the sensor network design problem. This problem amounts to find sets of variables to be measured by sensors for observation or control purposes as observability, disturbance rejection or diagnosis for example.
In this paper we address the sensor classification problem in the following sense. Given a system with its sensor network and a property P which is satisfied by this system with the existing sensors, we will classify the sensors with respect to their importance relatively to the preservation of the property P in case of failure. More precisely we will characterize the sensors which are critical, i.e. which failure leads to property loss and those which are useless for the property. We will also quantify the relative importance of the sensors which are neither useless nor critical.
We consider here linear structured system models, they represent a large class of parameter dependent linear systems. This allows us to use graph theory in order to easily exploit the structure of the process irrespective of the parameter values.
The properties which will be studied here are observability and Fault Detection and Isolation (FDI), we will then provide the sensor classification for these properties in case of sensor failure. The complexity of the classification algorithms is polynomial with respect to the dimension of the system.
The contribution of this approach is to provide with a unified framework allowing, with only a structural knowledge on the system, to determine which sensors are compulsory to use or useless to preserve a given property. Furthermore we propose a quantification of the respective importance of the useful sensors. The proposed graph approach is visual, easy to handle and close to the physical structure of the system. The underlying ideas are general and can be applied to other classes of models and properties.
back to top
Control in Semiconductor Wafer Manufacturing
*Abbas Emami-Naeini and Dick de Roover
*SC Solutions Inc. & Stanford University
emami@scsolutions.com
A semiconductor wafer undergoes a wide range of processes before it is transformed from a bare silicon wafer to one populated with millions of transistor circuits. Such processes include Physical or Chemical Vapor Deposition, (PVD, CVD), Chemical-Mechanical Planarization (CMP), Plasma Etch, Rapid Thermal Processing (RTP), and photolithography. As feature sizes keep shrinking, process control plays an increasingly important role in each of these processes. Model-based approach is an effective means of designing commercial controllers for advanced semiconductor equipment. We will give an overview of the applications of advanced control in the semiconductor industry. It is our experience that the best models for control design borrow heavily from the physics of the process. The manner in which these models are used for a specific control application depends on the performance goals. In some cases such as RTP and lithography, the closed-loop control depends entirely on having very good physical models of the system. For other processes such as CMP, physical models have to be combined with empirical models or are entirely empirical. The resulting multivariable controllers may be in-situ feedforward-feedback or run-to-run controllers, or a combination thereof. The three case studies that are presented in this paper (RTP, CMP, and lithography) are representative of the leading edge applications of advanced control in the semiconductor industry.
back to top
A Simple Polynomial Approach to Nonlinear Control
Mike J Grimble and Pawel Majecki*
Industrial Control Centre,
University of Strathclyde, UK
*ISC Limited, George Street, UK
m.grimble@eee.strath.ac.uk,
pawel@isc-ltd.com
http://www.icc.strath.ac.uk,
Although engineering systems are nonlinear most can be controlled adequately by linear control techniques. However, there is a class of systems for which linear control design methods are inadequate and the number of applications where this is occurring is increasing. This is due to increasingly stringent demands on the required performance of some high performance control systems. Such systems have to operate in zones where severe nonlinearities often dominate. In this case reliable and simple nonlinear control design methods are essential.
One of the simplest possible optimal control methods for linear systems is the so called generalized minimum variance controller on which the proposed nonlinear method builds. The main benefits of the so called NGMV approach lie in the simplicity of the concepts and in the straightforward structure of the algorithm. Since the introduction of NGMV designs a family of controllers has been introduced to deal with the different needs of stochastic or uncertain systems. These relate to their linear counterparts of factorised GMV, LQG, Generalised H8 and more recently generalized predictive controls.
In certain limiting cases, all of these algorithms revert to the basic NGMV design but in asymptotic cases for linear plant models the algorithms become identical to the well known linear versions. The polynomial operator versions of these algorithms are particularly easy to use and understand. The predictive controller is the most recent to be developed and seems to have great potential. The presentation will focus mostly on the predictive NGMV control which has been evaluated on reheat furnace controls and other potential applications.
back to top
Medical Robots and their Control Paradigms
Peter Kazanzides
Department of Computer Science,
Johns Hopkins University
pkaz@jhu.edu
http://www.cisst.org/~pkaz/
In this talk, I will provide an overview of my work in medical robotics, which began at IBM Research in 1989 with the development of the ROBODOC System for orthopaedic surgery and continued at a startup company that commercialized the technology. ROBODOC was used for over 20,000 hip and knee surgeries and recently received FDA approval. I will also describe several medical robots developed since I joined Johns Hopkins University in 2002.
The talk will then highlight the different high-level control paradigms employed in these systems, which vary from full automation to shared or cooperative control methods. In the medical environment, robot control must often balance the task goals with constraints imposed by the surgeon, the patient, or by intraoperative feedback from other devices. For example, the robot can create a virtual fixture to guide the surgeon or enforce a safety boundary; this fixture may need to be updated in real time to account for patient motion. The hope is that mechanical assistance, when coupled with appropriate control strategies, will enable surgeons to perform their tasks with greater safety and efficiency, thereby improving patient care.
back to top
Deadbeat Response is l2 Optimal
Vladimír Kucera
Director, Masaryk Institute of Advanced Studies
Faculty of Electrical Engineering
Czech Technical University in Prague, Czech Republic
kucera@muvs.cvut.cz
http://www.muvs.cvut.cz
A typical linear control strategy in discrete-time systems, deadbeat control produces transients that vanish in finite time. On the other hand, the linear-quadratic control stabilizes the system and minimizes the l2 norm of its transient response. Quite surprisingly, it is shown that deadbeat systems are l2 optimal, at least for reachable systems.
The proof makes use of polynomial matrix fractions and structure theorem for linear time-invariant multivariable systems, the notions introduced by W.A. Wolovich in the early seventies.
The result demonstrates the flexibility offered by the linear-quadratic regulator design and is an exercise in inverse optimality. The linear-quadratic regulator gain is unique, whereas the deadbeat feedback gains are not. Only one deadbeat gain is linear-quadratic optimal. An alternative construction of such a gain, based on solving an algebraic Riccati equation, is thus available.
back to top
Neural Network-Based Adaptive Optimal Controller - A Continuous-Time Formulation
D. Vrabie, F. Lewis, D. Levine*
Automation and Robotics Research Institute,
*Department of Psychology,
University of Texas at Arlington,
dvrabie@ uta.edu
http://arri.uta.edu/acs/
In this paper is presented new online adaptive control scheme, for partially unknown nonlinear systems, which converges to the optimal state-feedback control solution for affine in the input nonlinear systems. The main features of the algorithm map on the characteristics of the rewards-based decision making process in the mammal brain.
The derivation of the optimal adaptive control algorithm is presented in a continuous-time framework. The optimal control solution will be obtained in a direct fashion, without system identification. The algorithm is an online approach to policy iterations based on an adaptive critic structure to find an approximate solution to the state feedback, infinite-horizon, optimal control problem.
back to top
On Strongly Stabilizing Controller Synthesis
Hitay Ozbay
Dept. of Electrical and Electronics Eng.,
Bilkent University,
Ankara, Turkey
hitay@bilkent.edu.tr
http://www.ee.bilkent.edu.tr/~ozbay/
We present a small-gain based design method for synthesizing strongly stabilizing controllers (i.e. stable controllers stabilizing the feedback system) for various classes of linear time invariant plants. First, we consider multi-input-multi-output (MIMO) finite dimensional plants in our discussion. Next, extension of this technique to systems with time delays is illustrated. Furthermore, as a special case of stable controllers, we consider proportional plus derivative (PD) controller design for MIMO plants with input-output time delays.
back to top
From Centralized to Distributed Fault Detection: an Adaptive Approximation Approach
Thomas Parisini and *Marios M. Polycarpou
Dept. of Electrical, Electronic and Computer Engineering
University of Trieste
parisini@units.it
*Department of Electrical and Computer Engineering
University of Cyprus
mpolycar@ucy.ac.cy
This talk deals with the problem of designing a distributed fault detection methodology for distributed (and possibly large-scale) nonlinear dynamical systems. The adaptive approximation technique extends well-known results regarding nonlinear uncertain systems to the case of distributed nonlinear systems that are modelled as the interconnection of several subsystems. The subsystems are allowed to overlap, thus sharing some state components. For each subsystem, a Local Fault Detector is designed, based on the measured local state of the subsystem as well as the transmitted variables of neighboring states that define the subsystem interconnections. The local detection decision is made on the basis of the knowledge of the local subsystem dynamic model and of an adaptive approximation of the interconnection with neighboring subsystems. The use of a specially-designed consensus-based estimator is proposed in order to improve the detectability of faults affecting variables shared among different subsystems.
back to top
Rotational Motion with Almost Global Stability
Jose Vasconcelos, Anders Rantzer, Carlos Silvestre, Paulo Oliveira
Instituto Superior Técnico, Lisbon, Portugal
Anders Rantzer
Automatic Control LTH, Lund University, Sweden
anders.rantzer@control.lth.se
http://www.control.lth.se/~rantzer/
Global stability is usually a highly desirable property in schemes for control and estimation. However, for rotational motion there are topological constraints preventing solutions with fully global stability. The best one could hope for is an equilibrium that is "almost globally stable" in the sense that for all initial states except for a set of zero measure, the dynamics converge to the equilibrium. In this presentation, we show how Lyapunov functions can be combined with the so called density functions to prove the desirable stability property. In fact, combination of the two concepts turns out to be more powerful than any one of them on its own.
Results are illustrated on a special type of rotational dynamics where almost globally stability can be proved with density functions of a simple analytical form. In attitude estimation for rotational bodies, the argument can be extended to also prove that the stability property is robust to measurement noise.
back to top
Numerical Algorithms for Polynomial Plus/Minus Factorization
Martin Hromcik and Michael Sebek
Faculty of Electrical Engineering
Czech Technical University
Prague, Czech Republic
m.sebek@polyx.cz
www.michelsebek.cz
Two new algorithms are presented in the paper for the plus/minus factorization of a scalar discrete-time polynomial. The first method is based on the discrete Fourier transform theory (DFT) and its relationship to the Z-transform. Involving DFT computational techniques and the famous fast Fourier transform routine brings high computational efficiency and reliability. The method is applied in the case-study of $H_$2-optimal inverse dynamic filter to an audio equipment. The second numerical procedure originates in a symmetric spectral factorization routine, namely the Bauer's method of the 1950s. As a byproduct, a recursive LU factorization procedure for Toeplitz matrices is devised that is of more general impact and can be of use in other areas of applied mathematics as well. Performance of the method is demonstrated by an $l_1$ optimal controller design example.
back to top
Some Results on Input Classes for Identifiability
Eduardo Sontag
Rutgers University
New Brunswick, NJ, USA
This talk discusses what classes of input signals are sufficient in order to completely identify (in the absence of noise) the input/output behavior of generic bilinear systems. The main results are that step inputs are not sufficient, nor are single pulses, but the family of all pulses (of a fixed amplitude but varying widths) do suffice for identifiability. The work is joint with Yuan Wang and Alexandre Megretski.
back to top
Micromanipulation Using Vision and Force Feedback
Hakan Bilen and Mustafa Unel
Faculty of Engineering and Natural Sciences
Sabanci University, Istanbul Turkey
hakanbil@su.sabanciuniv.edu, munel@sabanciuniv.edu
http://people.sabanciuniv.edu/munel/
With the recent advances in the fields of micro and nanotechnology there has been growing interest in complex micromanipulation and microassembly strategies. Despite the fact that many commercially available micro devices such as the key components in automobile airbags, ink-jet printers and projection display systems are currently produced in a batch technique with little assembly, many other products such as read/write heads for hard disks and fiber optics assemblies require flexible precision assemblies. Furthermore, many biological micromanipulations such as invitro-fertilization, cell characterization and treatment rely on the ability of human operators. Requirement of high-precision, repeatable and financially viable operations in these tasks have given rise to the elimination of direct human involvement, and autonomy in micromanipulation and microassembly.
In this work, a flexible micromanipulation strategy based on vision and force feedback is developed. More specifically, a robust vision based control architecture is proposed and implemented to compensate errors due to uncertainties in position and shape of the microobjects to be manipulated. Furthermore, novel estimators are developed to identify the optical system and to characterize the mechanical properties of the biological structures through a synthesis of concepts from computer vision, estimation and control theory. Estimated mechanical parameters are utilized to reconstruct the imposed force on a biomembrane and to provide the adequate information to control the position, velocity and acceleration of the probe without damaging the cell/tissue during an injection task.
8:15-8:30 Welcome and Opening Remarks
8:30-10:30 Presentations
Deadbeat Response is l2 Optimal - Vladimir Kucera
A Simple Polynomial Approach to Nonlinear Control - Mike Grimble and Pawel Majecki
Numerical Algorithms for Polynomial Plus/Minus Factorization - Martin Hromcik and Michael Sebek
On Strongly Stabilizing Controller Synthesis - Hitay Özbay
Medical Robots and their Control Paradigms - Peter Kazanzides
Rotational Motion with Almost Global Stability - Jose Vasconcelos, Anders Rantzer, Carlos Silvestre, Paulo Oliveira
10:30-11:00 Coffee BreakA Simple Polynomial Approach to Nonlinear Control - Mike Grimble and Pawel Majecki
Numerical Algorithms for Polynomial Plus/Minus Factorization - Martin Hromcik and Michael Sebek
On Strongly Stabilizing Controller Synthesis - Hitay Özbay
Medical Robots and their Control Paradigms - Peter Kazanzides
Rotational Motion with Almost Global Stability - Jose Vasconcelos, Anders Rantzer, Carlos Silvestre, Paulo Oliveira
11:00-12:00 Presentations
Multivariable Zero-free Transfer Functions, and their Application in Economic Modeling - B. Anderson and M. Deistler
Zeros and Zero Dynamics for Linear, Time-delay Systems - G. Conte and A. M. Perdon
A Study of Feedback Fundamentals - Panos Antsaklis
12:00-12:30 Discussion - Then and Now in Control Zeros and Zero Dynamics for Linear, Time-delay Systems - G. Conte and A. M. Perdon
A Study of Feedback Fundamentals - Panos Antsaklis
Panos Antsaklis – Moderator
12:30-2:00 Lunch2:00-4:00 Presentations
Some Notes on Realization Theory - Shankar Bhattacharyya
Recent Advances in Positive Systems: Evolution to the Servo Problem - Bartek Roszak and Edward Davison
Neural Network-Based Adaptive Optimal Controller - A Continuous-Time Formulation - D. Vrabie, F. Lewis and D. Levine
Some Results on Input Classes for Identifiability - Eduardo D. Sontag
Control in Semiconductor Wafer Manufacturing - Abbas Emami-Naeini and Dick de Roover
Micromanipulation Using Vision and Force Feedback - Hakan Bilen and Mustafa Unel
4:00-4:30 Coffee BreakRecent Advances in Positive Systems: Evolution to the Servo Problem - Bartek Roszak and Edward Davison
Neural Network-Based Adaptive Optimal Controller - A Continuous-Time Formulation - D. Vrabie, F. Lewis and D. Levine
Some Results on Input Classes for Identifiability - Eduardo D. Sontag
Control in Semiconductor Wafer Manufacturing - Abbas Emami-Naeini and Dick de Roover
Micromanipulation Using Vision and Force Feedback - Hakan Bilen and Mustafa Unel
4:30-5:30 Presentations
New Trends in Neuro-Fuzzy Adaptive Control Schemes - Manolis A. Christodoulou and Yiannis S. Boutalis
Sensor Classification for Control and Diagnosis Problems, a Structural Approach - C. Commault, J.M. Dion and D.H. Trinh
From Centralized to Distributed Fault Detection: an Adaptive Approximation Approach - T. Parisini and M. Polycarpou
5:30-6:30 In Bill's HonorSensor Classification for Control and Diagnosis Problems, a Structural Approach - C. Commault, J.M. Dion and D.H. Trinh
From Centralized to Distributed Fault Detection: an Adaptive Approximation Approach - T. Parisini and M. Polycarpou
7:30 Dinner
Abstracts, Papers and Presentations (alphabetically by last name of speaker)
Multivariable Zero-free Transfer Functions, and their Application in Economic Modeling
Brian Anderson and Manfred Deistler
Abstract, Presentation
A Study of Feedback Fundamentals
Panos Antsaklis
Abstract, Presentation
Some Notes on Realization Theory
Shankar Bhattacharyya
Abstract, Presentation
New Trends in Neuro-Fuzzy Adaptive Control Schemes
Manolis A. Christodoulou and Yiannis S. Boutalis
Abstract, Paper, Presentation
Zeros and Zero Dynamics for Linear, Time-delay Systems
G. Conte and A. M. Perdon
Abstract, Paper, Presentation
Recent Advances in Positive Systems: Evolution to the Servo Problem
Bartek Roszak and Edward Davison
Abstract, Presentation
Sensor Classification for Control and Diagnosis Problems, a Structural Approach
C. Commault, J.M. Dion and D.H. Trinh
Abstract, Paper, Presentation
Control in Semiconductor Wafer Manufacturing
Abbas Emami-Naeini and Dick de Roover
Abstract, Paper
A Simple Polynomial Approach to Nonlinear Control
Mike Grimble and Pawel Majecki
Abstract, Presentation
Medical Robots and their Control Paradigms
Peter Kazanzides
Abstract, Presentation
Deadbeat Response is l2 Optimal
Vladimir Kucera
Abstract, Paper, Presentation
Neural Network-Based Adaptive Optimal Controller - A Continuous-Time Formulation
D. Vrabie, F. Lewis and D. Levine
Abstract, Paper, Presentation
On Strongly Stabilizing Controller Synthesis
Hitay Özbay
Abstract, Presentation
From Centralized to Distributed Fault Detection: an Adaptive Approximation Approach
Thomas Parisini and Marios M. Polycarpou
Abstract, Presentation
Rotational Motion with Almost Global Stability
Jose Vasconcelos, Anders Rantzer, Carlos Silvestre, Paulo Oliveira
Abstract, Presentation
Numerical Algorithms for Polynomial Plus/Minus Factorization
Martin Hromcik and Michael Sebek
Abstract, Paper, Presentation
Some Results on Input Classes for Identifiability
Eduardo D. Sontag
Abstract, Presentation
Micromanipulation Using Vision and Force Feedback
Hakan Bilen and Mustafa Unel
Abstract, Presentation
Abstracts and Contact Information for the above Speakers and Titles
Multivariable Zero-free Transfer Functions, and their Application in Economic Modeling
Brian D O Anderson, Australian National University
Manfred Deistler, Technical University of Vienna
brian.anderson@anu.edu.au, deistler@tuwien.ac.at
http://users.rsise.anu.edu.au/~briandoa/
http://www.eos.tuwien.ac.at/Oeko/MDeistler/
Central banks and funds investment managers work with mathematical models. In recent years, a new class of model has come into prominence—generalized dynamic factor models. These are characterized by having a modest number of inputs, corresponding to key economic variables and industry-sector-wide variables for central banks and funds managers respectively, and a large number of outputs, economic time series data or individual stock price movements for example. It is common to postulate that the input variables are linked to the output variables by a finite-dimensional linear discrete-time dynamic model, the outputs of which are corrupted by noise to yield the measured data. The key problems faced by central banks or funds managers are model fitting given the output data (but not the input data), and using the model for prediction purposes.
These are essentially tasks usually considered by those practicing identification and time series modelling. Nevertheless there is considerable underlying linear system theory. This flows from the fact that the underlying transfer function matrix is tall.
This talk will describe a number of consequences of this seemingly trivial fact. For example, a tall transfer function of known McMillan degree but otherwise generic has no zeros, finite or infinite. A finite sequence of output data in the discrete time case allows recovery of a finite sequence of input data, without knowledge of the initial state. Canonical state-variable forms take on a special format.
Key references outlining this work are as follows: Anderson, B.D.O., Deistler, M., Generalized linear dynamic factor models—a structure theory, CDC 2008, to appear Anderson, B.D.O. and Deistler, M., Properties of zero-free transfer function matrices. SICE Journal of Control, Measurement and System Integration, to appear
back to top
A Study of Feedback Fundamentals
Panos Antsaklis
Department of Electrical Engineering, University of Notre Dame
antsaklis.1@nd.edu
http://www.nd.edu/~pantsakl
In the area of Systems and Control theory, the emphasis has been on designing feedback controllers given a model of the process to be controlled and a number of control specifications. Many powerful methodologies have been introduced in the past half century to design controllers that stabilize and achieve desired performance in a robust way, being tolerant to certain class of plant parameter variations and external disturbances. Feedback or closed loop control, instead of feed-forward or open loop control, is used because of uncertainties in the plant and its environment. Significantly less effort has been spent in the past half century on understanding exactly how and why feedback works so well not only in the control of engineered systems but in natural systems as well. What are the fundamental principles, the fundamental mechanisms, which make feedback control so powerfully effective? These fundamental mechanisms should be independent of the particular type of mathematical models used, that is the system may be described by differential equations, by automata, by logic expressions, by natural language since we do know that feedback is ubiquitous and works! What are these fundamental properties that are present everywhere? This talk will raise many questions and I hope will provide a few answers as well.
back to top
Some Notes on Realization Theory
S.P.Bhattacharyya
Dept. of Electrical and Computer Engg.,
Texas A & M University
bhatt@ece.tamu.edu
http://www.ece.tamu.edu/People/bios/bbhattac.html
In this talk, prepared in honor of Prof. Wolovich’s 70th birthday, we discuss some issues related to minimal and non-minimal realizations and their role in controller synthesis. In particular, it is pointed out that realizations may be generically minimal or fragile with respect to system parameters. The latter class of systems are minimal only on an algebraic variety. Reliable control and stabilization can be achieved in such cases by first perturbing the system away from this variety to restore its locally maximal order and to base controller design based on this perturbed model. Examples illustrating this departure from the conventional wisdom of always using minimal realizations, will be included in the talk.
back to top
New Trends in Neuro-Fuzzy Adaptive Control Schemes
Manolis A. Christodoulou and Yiannis S. Boutalis*
Department of Electronic and Computer Engineering
Technical University of Crete, Greece
*Department of Electrical and Computer Engineering
University of Thraki, Greece & University of Erlangen-Nuremberg, Germany
manolis@ece.tuc.gr
ybout@ee.duth.gr
We use a new definition of Neuro-Fuzzy Dynamical Systems, using the concept of Fuzzy Dynamical Systems (FDS) in conjunction with High Order Neural Network Functions (F-HONNFs). The dynamical System is assumed nonlinear and totally unknown. We first propose its approximation by a special form of a fuzzy dynamical system (FDS) and in the sequel the fuzzy rules are approximated by appropriate HONNF’s. Thus the identification scheme leads to a Recurrent High Order Neural Network, which however, takes into account the fuzzy output partitions of the initial FDS. The proposed scheme does not require a priori experts’ information on the number and type of input variable membership functions, making it less vulnerable to initial design assumptions.
After the identification process we adaptively control the system either directly or indirectly. By doing so, we present weight updating laws for the involved HONNs. With rigorous proofs we guarantee that the errors converge to zero exponentially fast, or at least become uniformly ultimately bounded. At the same time we guarantee stability by proving that all signals in the closed loop remain bounded.
During both the identification and control process we assume, first that we know the centers and shapes of membership functions, and we identify the HONN parameters in which case we get a directional variation. Thus in order to guarantee existence of the control law, we define a new method replacing the well known projection, which is termed parameter hopping and thus we rigorously prove existence of the control law, guaranteeing stability properties.
In the sequel we also assume, that both membership function centers and HONN parameters are identifiable, ending up with a bilinear multivariable adaptive control problem which is solved for the first time in the adaptive controls literature. In this last case we are able to adaptively controlling the system, while the updating laws automatically update the parameters as well as the centers of the membership functions, placing them optimally in space. The only requirement is that we know the signs of the centers, a condition that may be relaxed in our future research work using a method similar to the Nussbaum gain for scalar-vector bilinear cases.
Simulations illustrate the potency of the method and comparisons with conventional approaches are given. The simulation tests are based on benchmark examples. Also, the applicability of the method is tested on a DC Motor system where it is shown that by following the proposed procedure one can obtain asymptotic regulation.
back to top
Zeros and Zero Dynamics for Linear, Time-delay Systems
G. Conte and A. M. Perdon
Dipartimento di Ingegneria Informatica, Gestionale e dell’Automazione
Università Politecnica delle Marche, Ancona , Italy
gconte@univpm.it, perdon@univpm.it
The notion of zero of a linear, dynamical system can be investigated and studied from many different points of view. In particular, the algebraic approach proposed some time ago by M. Sain and B. Wyman provides conceptual and practical tools for generalizing the notion of zero to several classes of dynamical systems, notably to systems with coefficients in a ring. By representing time delay systems as systems with coefficients in a ring, it is possible to introduce a suitable notion of zeros and of zero dynamics in the time delay framework, that turns out to be useful for studying interesting control problems. In this paper, we discuss such notions and their features, both from an algebraic point of view and from a geometric one, highlighting the relationship between zeros and controlled invariance subspaces of the state space. In addition, we investigate the role of zeros and zero dynamics in the design of feedback control loops and in inversion problems involving time delay systems and we show the practical use of those notions in dealing with tracking problems in the time-delay framework.
back to top
Recent Advances in Positive Systems: Evolution to the Servo Problem
Bartek Roszak and Edward J Davison*
Dept of Electrical and Computer Engineering
University of Toronto
ted@control.utoronto.ca
http://www.control.toronto.edu/people/profs/ted/ted.html
The servomechanism problem has been of importance in the emergence of LTI systems over the span of several decades. However for the class of “positive systems” , there is a gap in the research literature in that the tracking and disturbance rejection problem, although of great importance, has been ignored .Thus the goal of this talk is to give an overview of recent results obtained for the tracking and disturbance problem for positive LTI systems.
A positive LTI system is an LTI system with the imposed constraints that the state, output and/or input variables be nonnegative for all time. The motivation for studying these systems is that the nonnegative property occurs quite frequently in various applications and in nature, e.g. they immediately arise in hydrology, engineering stocking, baking ovens, furnace systems, building temperature control systems, in almost all areas of biology ….
The talk will present various results discussing the servomechanism problem under measurable or unmeasurable disturbances for both known (or unknown) positive LTI systems.
back to top
Sensor Classification for Control and Diagnosis Problems, a Structural Approach
C. Commault, J.M. Dion and D.H. Trinh
GIPSA-Lab Grenoble
Department of automatic control
Institut Polytechnique de Grenoble, France
jean-michel.dion@gipsa-lab.inpg.fr
http://www.gipsa-lab.inpg.fr/
Control and diagnosis of dynamical systems require measures or estimation of system variables via sensors. In this paper we consider the sensor network design problem. This problem amounts to find sets of variables to be measured by sensors for observation or control purposes as observability, disturbance rejection or diagnosis for example.
In this paper we address the sensor classification problem in the following sense. Given a system with its sensor network and a property P which is satisfied by this system with the existing sensors, we will classify the sensors with respect to their importance relatively to the preservation of the property P in case of failure. More precisely we will characterize the sensors which are critical, i.e. which failure leads to property loss and those which are useless for the property. We will also quantify the relative importance of the sensors which are neither useless nor critical.
We consider here linear structured system models, they represent a large class of parameter dependent linear systems. This allows us to use graph theory in order to easily exploit the structure of the process irrespective of the parameter values.
The properties which will be studied here are observability and Fault Detection and Isolation (FDI), we will then provide the sensor classification for these properties in case of sensor failure. The complexity of the classification algorithms is polynomial with respect to the dimension of the system.
The contribution of this approach is to provide with a unified framework allowing, with only a structural knowledge on the system, to determine which sensors are compulsory to use or useless to preserve a given property. Furthermore we propose a quantification of the respective importance of the useful sensors. The proposed graph approach is visual, easy to handle and close to the physical structure of the system. The underlying ideas are general and can be applied to other classes of models and properties.
back to top
Control in Semiconductor Wafer Manufacturing
*Abbas Emami-Naeini and Dick de Roover
*SC Solutions Inc. & Stanford University
emami@scsolutions.com
A semiconductor wafer undergoes a wide range of processes before it is transformed from a bare silicon wafer to one populated with millions of transistor circuits. Such processes include Physical or Chemical Vapor Deposition, (PVD, CVD), Chemical-Mechanical Planarization (CMP), Plasma Etch, Rapid Thermal Processing (RTP), and photolithography. As feature sizes keep shrinking, process control plays an increasingly important role in each of these processes. Model-based approach is an effective means of designing commercial controllers for advanced semiconductor equipment. We will give an overview of the applications of advanced control in the semiconductor industry. It is our experience that the best models for control design borrow heavily from the physics of the process. The manner in which these models are used for a specific control application depends on the performance goals. In some cases such as RTP and lithography, the closed-loop control depends entirely on having very good physical models of the system. For other processes such as CMP, physical models have to be combined with empirical models or are entirely empirical. The resulting multivariable controllers may be in-situ feedforward-feedback or run-to-run controllers, or a combination thereof. The three case studies that are presented in this paper (RTP, CMP, and lithography) are representative of the leading edge applications of advanced control in the semiconductor industry.
back to top
A Simple Polynomial Approach to Nonlinear Control
Mike J Grimble and Pawel Majecki*
Industrial Control Centre,
University of Strathclyde, UK
*ISC Limited, George Street, UK
m.grimble@eee.strath.ac.uk,
pawel@isc-ltd.com
http://www.icc.strath.ac.uk,
Although engineering systems are nonlinear most can be controlled adequately by linear control techniques. However, there is a class of systems for which linear control design methods are inadequate and the number of applications where this is occurring is increasing. This is due to increasingly stringent demands on the required performance of some high performance control systems. Such systems have to operate in zones where severe nonlinearities often dominate. In this case reliable and simple nonlinear control design methods are essential.
One of the simplest possible optimal control methods for linear systems is the so called generalized minimum variance controller on which the proposed nonlinear method builds. The main benefits of the so called NGMV approach lie in the simplicity of the concepts and in the straightforward structure of the algorithm. Since the introduction of NGMV designs a family of controllers has been introduced to deal with the different needs of stochastic or uncertain systems. These relate to their linear counterparts of factorised GMV, LQG, Generalised H8 and more recently generalized predictive controls.
In certain limiting cases, all of these algorithms revert to the basic NGMV design but in asymptotic cases for linear plant models the algorithms become identical to the well known linear versions. The polynomial operator versions of these algorithms are particularly easy to use and understand. The predictive controller is the most recent to be developed and seems to have great potential. The presentation will focus mostly on the predictive NGMV control which has been evaluated on reheat furnace controls and other potential applications.
back to top
Medical Robots and their Control Paradigms
Peter Kazanzides
Department of Computer Science,
Johns Hopkins University
pkaz@jhu.edu
http://www.cisst.org/~pkaz/
In this talk, I will provide an overview of my work in medical robotics, which began at IBM Research in 1989 with the development of the ROBODOC System for orthopaedic surgery and continued at a startup company that commercialized the technology. ROBODOC was used for over 20,000 hip and knee surgeries and recently received FDA approval. I will also describe several medical robots developed since I joined Johns Hopkins University in 2002.
The talk will then highlight the different high-level control paradigms employed in these systems, which vary from full automation to shared or cooperative control methods. In the medical environment, robot control must often balance the task goals with constraints imposed by the surgeon, the patient, or by intraoperative feedback from other devices. For example, the robot can create a virtual fixture to guide the surgeon or enforce a safety boundary; this fixture may need to be updated in real time to account for patient motion. The hope is that mechanical assistance, when coupled with appropriate control strategies, will enable surgeons to perform their tasks with greater safety and efficiency, thereby improving patient care.
back to top
Deadbeat Response is l2 Optimal
Vladimír Kucera
Director, Masaryk Institute of Advanced Studies
Faculty of Electrical Engineering
Czech Technical University in Prague, Czech Republic
kucera@muvs.cvut.cz
http://www.muvs.cvut.cz
A typical linear control strategy in discrete-time systems, deadbeat control produces transients that vanish in finite time. On the other hand, the linear-quadratic control stabilizes the system and minimizes the l2 norm of its transient response. Quite surprisingly, it is shown that deadbeat systems are l2 optimal, at least for reachable systems.
The proof makes use of polynomial matrix fractions and structure theorem for linear time-invariant multivariable systems, the notions introduced by W.A. Wolovich in the early seventies.
The result demonstrates the flexibility offered by the linear-quadratic regulator design and is an exercise in inverse optimality. The linear-quadratic regulator gain is unique, whereas the deadbeat feedback gains are not. Only one deadbeat gain is linear-quadratic optimal. An alternative construction of such a gain, based on solving an algebraic Riccati equation, is thus available.
back to top
Neural Network-Based Adaptive Optimal Controller - A Continuous-Time Formulation
D. Vrabie, F. Lewis, D. Levine*
Automation and Robotics Research Institute,
*Department of Psychology,
University of Texas at Arlington,
dvrabie@ uta.edu
http://arri.uta.edu/acs/
In this paper is presented new online adaptive control scheme, for partially unknown nonlinear systems, which converges to the optimal state-feedback control solution for affine in the input nonlinear systems. The main features of the algorithm map on the characteristics of the rewards-based decision making process in the mammal brain.
The derivation of the optimal adaptive control algorithm is presented in a continuous-time framework. The optimal control solution will be obtained in a direct fashion, without system identification. The algorithm is an online approach to policy iterations based on an adaptive critic structure to find an approximate solution to the state feedback, infinite-horizon, optimal control problem.
back to top
On Strongly Stabilizing Controller Synthesis
Hitay Ozbay
Dept. of Electrical and Electronics Eng.,
Bilkent University,
Ankara, Turkey
hitay@bilkent.edu.tr
http://www.ee.bilkent.edu.tr/~ozbay/
We present a small-gain based design method for synthesizing strongly stabilizing controllers (i.e. stable controllers stabilizing the feedback system) for various classes of linear time invariant plants. First, we consider multi-input-multi-output (MIMO) finite dimensional plants in our discussion. Next, extension of this technique to systems with time delays is illustrated. Furthermore, as a special case of stable controllers, we consider proportional plus derivative (PD) controller design for MIMO plants with input-output time delays.
back to top
From Centralized to Distributed Fault Detection: an Adaptive Approximation Approach
Thomas Parisini and *Marios M. Polycarpou
Dept. of Electrical, Electronic and Computer Engineering
University of Trieste
parisini@units.it
*Department of Electrical and Computer Engineering
University of Cyprus
mpolycar@ucy.ac.cy
This talk deals with the problem of designing a distributed fault detection methodology for distributed (and possibly large-scale) nonlinear dynamical systems. The adaptive approximation technique extends well-known results regarding nonlinear uncertain systems to the case of distributed nonlinear systems that are modelled as the interconnection of several subsystems. The subsystems are allowed to overlap, thus sharing some state components. For each subsystem, a Local Fault Detector is designed, based on the measured local state of the subsystem as well as the transmitted variables of neighboring states that define the subsystem interconnections. The local detection decision is made on the basis of the knowledge of the local subsystem dynamic model and of an adaptive approximation of the interconnection with neighboring subsystems. The use of a specially-designed consensus-based estimator is proposed in order to improve the detectability of faults affecting variables shared among different subsystems.
back to top
Rotational Motion with Almost Global Stability
Jose Vasconcelos, Anders Rantzer, Carlos Silvestre, Paulo Oliveira
Instituto Superior Técnico, Lisbon, Portugal
Anders Rantzer
Automatic Control LTH, Lund University, Sweden
anders.rantzer@control.lth.se
http://www.control.lth.se/~rantzer/
Global stability is usually a highly desirable property in schemes for control and estimation. However, for rotational motion there are topological constraints preventing solutions with fully global stability. The best one could hope for is an equilibrium that is "almost globally stable" in the sense that for all initial states except for a set of zero measure, the dynamics converge to the equilibrium. In this presentation, we show how Lyapunov functions can be combined with the so called density functions to prove the desirable stability property. In fact, combination of the two concepts turns out to be more powerful than any one of them on its own.
Results are illustrated on a special type of rotational dynamics where almost globally stability can be proved with density functions of a simple analytical form. In attitude estimation for rotational bodies, the argument can be extended to also prove that the stability property is robust to measurement noise.
back to top
Numerical Algorithms for Polynomial Plus/Minus Factorization
Martin Hromcik and Michael Sebek
Faculty of Electrical Engineering
Czech Technical University
Prague, Czech Republic
m.sebek@polyx.cz
www.michelsebek.cz
Two new algorithms are presented in the paper for the plus/minus factorization of a scalar discrete-time polynomial. The first method is based on the discrete Fourier transform theory (DFT) and its relationship to the Z-transform. Involving DFT computational techniques and the famous fast Fourier transform routine brings high computational efficiency and reliability. The method is applied in the case-study of $H_$2-optimal inverse dynamic filter to an audio equipment. The second numerical procedure originates in a symmetric spectral factorization routine, namely the Bauer's method of the 1950s. As a byproduct, a recursive LU factorization procedure for Toeplitz matrices is devised that is of more general impact and can be of use in other areas of applied mathematics as well. Performance of the method is demonstrated by an $l_1$ optimal controller design example.
back to top
Some Results on Input Classes for Identifiability
Eduardo Sontag
Rutgers University
New Brunswick, NJ, USA
This talk discusses what classes of input signals are sufficient in order to completely identify (in the absence of noise) the input/output behavior of generic bilinear systems. The main results are that step inputs are not sufficient, nor are single pulses, but the family of all pulses (of a fixed amplitude but varying widths) do suffice for identifiability. The work is joint with Yuan Wang and Alexandre Megretski.
back to top
Micromanipulation Using Vision and Force Feedback
Hakan Bilen and Mustafa Unel
Faculty of Engineering and Natural Sciences
Sabanci University, Istanbul Turkey
hakanbil@su.sabanciuniv.edu, munel@sabanciuniv.edu
http://people.sabanciuniv.edu/munel/
With the recent advances in the fields of micro and nanotechnology there has been growing interest in complex micromanipulation and microassembly strategies. Despite the fact that many commercially available micro devices such as the key components in automobile airbags, ink-jet printers and projection display systems are currently produced in a batch technique with little assembly, many other products such as read/write heads for hard disks and fiber optics assemblies require flexible precision assemblies. Furthermore, many biological micromanipulations such as invitro-fertilization, cell characterization and treatment rely on the ability of human operators. Requirement of high-precision, repeatable and financially viable operations in these tasks have given rise to the elimination of direct human involvement, and autonomy in micromanipulation and microassembly.
In this work, a flexible micromanipulation strategy based on vision and force feedback is developed. More specifically, a robust vision based control architecture is proposed and implemented to compensate errors due to uncertainties in position and shape of the microobjects to be manipulated. Furthermore, novel estimators are developed to identify the optical system and to characterize the mechanical properties of the biological structures through a synthesis of concepts from computer vision, estimation and control theory. Estimated mechanical parameters are utilized to reconstruct the imposed force on a biomembrane and to provide the adequate information to control the position, velocity and acceleration of the probe without damaging the cell/tissue during an injection task.