Partenaires

Ampère

Nos tutelles

CNRS Ecole Centrale de Lyon Université de Lyon Université Lyon 1 INSA de Lyon

Nos partenaires

Ingénierie@Lyon



Rechercher


Accueil > Actualités

Séminaires scientifiques

par Laurent Krähenbühl - publié le , mis à jour le

Agenda

Ajouter un événement iCal

Liens vers les séminaires de 2019, 2018, 2017, 2016, 2015, 2014, 2013, 2012, 2011, 2010.


Prochains séminaires :


Ci-dessous la liste des derniers séminaires :

15/02/2019, 12h30-14h00 : Alexandre Kircher (Ampère) - PhD Club


ECL, H9, salle TIC

Resilient State Estimation for Linear Time-Varying Discrete Systems
In the past sixty years, most of the state estimators have been developed to handle disturbances modeled as random signals with known density functions (e.g. gaussian noises). However, new challenges, inspired by architectures of systems such as the Cyber-Physical Systems (CPS), have appeared in order to design estimators capable of coping with other models of disturbances. This presentation will introduce an optimisation-based approach to estimate systems in presence of sparse noise . Not falling in the category handled by classic state estimators, this type of noise is almost always equal to zero but its non-zero elements can arbitrarily take any value, which represents many diverse phenomenoms such as sensor failures, transmission delays or even attacks on the system. To assess the viability of this new structure of estimation, a particular case of estimator will be studied in two configurations : one where only the sparse noise is present and one where a process bounded noise is also present. In the first case, an equivalent condition to ensure the exact recoverability of the system (i.e. the estimation process exactly yields the system state) will be displayed. In the second case, we can find an upper bound to the norm of the estimation error which is independent from the values the sparse noise takes. Along those theoretical properties, a few simulation results will shed light upon this estimation structure.

14/02/2019, 14h00-15h30 : Daniele Astolfi (LAGEP)


INSA, Saint-Exupéry, 1er étage, salle de réunion Ampère

The challenges of nonlinear robust output regulation
This talk deals with the problem of output regulation for nonlinear systems. In the first part of the talk, we present some key notions and tools governing the design of internal model-based regulators for systems. Then, the case of nonlinear systems is discussed. We overview the main results presented in literature in the last decade for single-input single-error systems, with a special emphasis toward high gain-based methodologies. Robustness of these solutions is discussed. Finally, in the last part of the talk, preliminary new results of robust output regulation for specific classes of nonlinear systems are presented.

12/02/2019, 10h00-12h00 : Laurentiu Hetel et Alexandre Kruszewski (CRISTAL, Lille)

INSA, Saint-Exupéry, 1er étage, salle de réunion Ampère.

Nos deux invités viennent du Centre de Recherche en Informatique, Signal et Automatique de Lille (CRISTAL, CNRS UMR9189)

About the stabilization of switched affine systems
L. Hetel est CR (HDR) dans l’équipe SHOC.
In this presentation we considers the stabilization problem for the class of switched affine systems. The main idea is to use an alternative representation of the switched affine system as a nonlinear system with input constraint in a finite alphabet. Switching laws can be derived by emulating locally classical continuous controllers. It is shown that the classical constrainton the existence of constant stable convex combinations may be easily avoided.The approach may be interpreted as a generalization where convex combinations defined as functions of the system state are being used. Constructive methods for deriving switching laws are proposed.
Commande basée modèle pour les robots déformables
A. Kruszewski est MCF (HDR) dans l’équipe DEFROST.
La robotique déformable pourrait constituer la prochaine révolution dans le domaine de la robotique. Plusieurs verrous scientifiques importants limitent leur développement notamment au niveau de la modélisation et de la commande. Dans cette présentation, j’aborderai les problématiques étudiées par l’équipe DEFROST (CRIStAL - INRIA) en lien avec la commande de ces systèmes. A partir des modèles éléments finis de ces robots et en s’appuyant sur des méthodes d’optimisation, de réduction de modèle ainsi que les méthodes de Lyapunov, des lois de commande sont conçues pour améliorer le comportement dynamique et la précision.