ICS RAS L.7 V.A. Trapeznikov Institute of Control Sciences of Russian Academy of Sciences
Laboratory 7
Adaptive and Robust Systems

Last modified December 1, 2011.

Seminars

Laboratory 7 of the Institute of Control Sciences RAS organizes a series of seminars
on the theory of automatic control and optimization.
Seminars are taking place every Tuesday from 11:30 in room 433.
Address:  Institute of Control Sciences (4th floor), 65 Profsoyuznaya str., Moscow.

Seminar chair:  

B. Polyak,  boris ipu.ru

Seminar secretariat:  

Elena Stefanyuk,  stefa ipu.ru
Pavel Shcherbakov,  sherba ipu.ru
Phone: (+7-495) 334-76-41

Only talks in English are presented. For full list, including archives, please refer to Russian version.
There also exists frequently updated mirror (in Russian too).


Schedule (2012)

No English talks scheduled.

Archive

2012

June 26Stephen Boyd (Stanford University, USA)
"Distributed Optimization via the Alternating Direction Method of Multipliers"

Slides, paper and examples are at special page http://www.stanford.edu/~boyd/papers/admm_distr_stats.html.


June 26Noboru Sakamoto (Nagoya University, Japan)
"Nonlinear output regulation theory based on center-stable manifold computation" (slides, video - two swings, video - three swings)

In this talk, first, invariant manifold theory, especially, the theory of center-stable manifold is reviewed, and next, it will be explained how this theory is essentia to design nonlinear output regulators without solving the nonlinear regulator equation. This regulators are optimal with respect to a standard cost function and its computation is accomplished in the framework of the computation of a solution of Hamilton-Jacobi equations. Several numerical examples show its effectiveness and potentials for industry applications.


November 27 (11:30, r. 433)Benzerrouk H. (Saint Petersburg State University of Aerospace Instrumentation)
"Robust Non Linear Filters Applied to Original Problems in Aerospace Integrated Navigation Systems" (slides)

In this work, deep analysis of modern non linear filtering methods is presented including Extended Kalman Filter EKF, Sigma Point Kalman Filters SPKF, Divided Differences Kalman Filters DDF and the recently developed Cubature based Kalman Filter CKF. These algorithms are applied to UAV autonomous navigation based on INS/GNSS data fusion. The direct non linear filtering approach is selected instead the indirect filtering, which involves linearization of the non linear kinematic model of the vehicle. In order to provide realistic results, adaptive forms of these analysed algorithms are developed based on innovation adaptation and sub-optimal adaptive fading factor. On the other hand, and in order to lead with real measurement noise environment, robust variants based on Gaussian mixture filtering are investigated and tested in simulation. Thus, adaptive and robust non linear filtering is proposed as an alternative to time varying non Gaussian measurement noises. Finally, in order to validate the developed approaches, three applications with non linear system differential equations are carried out; 2D Car navigation based on IMU/GNSS/Compass integrated system in denied GNSS environment, Pedestrian Navigation System for blinds peoples in the city based on foot-mounted IMU and Zero velocity update techniques and original integrated navigation system on the surface of Mars planet for astronauts and rovers navigation. Results are discussed and the CKF algorithm is selected for such non linear filtering problems.


2011

December 6 (14:00, r. 433) — Professor and head of Bio-medical Engineering Department Mingyue Ding (Huazhong University of Science and Technology, Wuhan, China)
"Advances in Medical Ultrasound Laboratory of Huazhong University of Science and Technology" (slides)

In this talk, first, an introduction of the MUL was given including the year built and the people of the lab as well as its mission. Secondly, the main research areas of the lab were described and the faculty and students involved are listed. After that, the main advances in the MUL including the research projects, publication, awards as well as 3D ultrasound imaging commercialization are discussed. Finally, the future works of research interests are addressed.

"3D Ultrasound imaging and its medical applications"
First, what is 3D ultrasound imaging technology, why we need to develop a 3D ultrasound imaging technology were discussed. Then, the different techniques of 3D ultrasound imaging were introduced. After that, the main applications in medicine were given, including diseases diagnosis, image-guided interventional surgery and therapy and the evaluation of therapy and surgery and follow up. Finally, the researches in the Medical Ultrasound Laboratory of Huazhong University of Science and Technology in China and main advances of the research were described. Furthermore, the future works are given.


November, 22 — RAS member (corr.) G.A. Leonov (Saint-Petersburg State University, Russia)
"Hidden oscillations in nonlinear dynamical systems" (slides)


2010

January, 19M.B. Malyutov (Northeastern University, Boston, MA, USA)
"Recovery of Sparse Active Inputs in general systems" (slides)

Around 600 papers were published last 5 years on COMPRESSIVE SENSING after Donoho (genius of advertising) et al applied practical L1 optimization (see e.g. www.dsp.ece.rice.edu/cs). Multi-Access Information Theory based applications obtained since 1979 can greatly enhance the current understanding of sparse recovery. Some of the alternative competitive practical algorithms known since 1959 outperform L1 optimization in a wide range of models are applicable to general nonparametric noisy models including "noise with memory".


April 20 — (talk was cancelled) Dr. Alex Poznyak (Dept. Control Automatico, CINVESTAV-IPN, Mexico)
"Robust Nonlinear Control for Implicit System Using Attractive Ellipsoid Technique"


2009

September, 15A. Poznyak (CINVESTAV, Mexico)
"Robust Nonlinear Sample-Data Feedback Control: Attracting Ellipsoid Method"

This talk addresses the problem of robust control for a class of nonlinear dynamical systems in the discrete-continuous time domain. We deal with nonlinear controllable models described by ordinary differential equations in the presence of bounded uncertainties. The full model of the control system under consideration is completed by linear sampling-type outputs. The linear feedback control design proposed in this manuscript is created by application of an extended version of the conventional attracting ellipsoid method. Moreover, we also apply a new Lyapunov-based "descriptor techniques" (some analogue of Lagrange multipliers for Banach spaces). The above combination of the modified attracting ellipsoid approach and descriptor method make it possible to obtain the robustness of the designed control and to establish some well stability properties of the considered dynamical systems. Finally, the applicability of the proposed method is illustrated by a computational example. A brief discussion on the main implementation issue will be also included.


April 21Romeo Ortega (Laboratoire des Signaux et Systemes, Supelec, Gif-sur-Yvette, France)
"Control by Interconnection of Port-Hamiltonian Systems" (slides)


July 13Shinji Hara (Dept. of Information Physics and Computing, The University of Tokyo, Japan)
"Consensus and Cooperation in Multi-Agent Dynamical Systems: A Unified Approach Based on Systems with Generalized Frequency Variables and Its Applications" (slides)


July 16Dimitri Peaucelle (LAAS, Toulouse, France)
"Robust Performance Analysis in Quadratic Separation Framework" (slides)


2008

January 29V. Sokolov (Komi Science Center, RAS)
"Suboptimal robust control of unknown first order plant with delay" (slides)


April 1S. Nazin (ICS RAS, Moscow)
"3-rd Order Newton Method and its Global Performance" (slides)


June 5A. Poznyak (CINVESTAV, Mexico)
"Output Linear Robust Stabilization for a Class of Nonlinear Systems" (slides)


2007

October 9Fabrizio Dabbene (Politecnico di Torino, Italy)
"A Probabilistic Analytic Center Cutting Plane Method for Feasibility of Uncertain LMIs" (slides)

Many robust control problems can be formulated in abstract form as convex feasibility programs, where one seeks a solution vector x that satisfies a set of inequalities of the form F = { f(x,δ) ≤ 0,  δ from D }. This set typically contains an infinite and uncountable number of inequalities, and it has been proved that the related robust feasibility problem is numerically hard to solve in general. In this talk, we discuss a family of cutting plane methods that solve efficiently a probabilistically-relaxed version of the problem. Specifically, under suitable hypotheses, we show that an Analytic Center Cutting Plane scheme based on a probabilistic oracle returns in a finite and pre-specified number of iterations a solution x which is feasible for most of the members of F, except possibly for a subset having arbitrarily small probability measure.



L7 2003—2016