| Libellé du cours : | Real time estimation for engineers |
|---|---|
| Département d'enseignement : | EEA / Electronique Electrotechnique Automatique |
| Responsable d'enseignement : | Monsieur WILFRID PERRUQUETTI |
| Langue d'enseignement : | Français |
| Ects potentiels : | 4 |
| Grille des résultats : | Grade de A+ à R |
| Code et libellé (hp) : | G1G2_ED_EEA_RTE - Real time estimat.engineers |
Equipe pédagogique
Enseignants : Monsieur WILFRID PERRUQUETTI
Intervenants extérieurs (entreprise, recherche, enseignement secondaire) : divers enseignants vacataires
Résumé
For engineers, a wide variety of information cannot be directly obtained through measurements. Some parameters (constants of an electrical actuator, delay in a transmission, ….) or internal variables (robot’s posture, torques applied to a robot, …) are unknown or are not measured by physical sensors. Thus, it is needed to extract the information conveyed by the signals in order to estimate (when possible) the missing information. These “real-time software sensors” can be viewed as cheap but efficient tools allowing one to recover variables of interest from available measurements. Estimation has a long-standing history going back to earlier mathematics. It consists in finding some quantities from measured ones. The quantities to be found may be related to a statistic model, a static model or a dynamic one (Ordinary Differential Equations, Partial Differential Equations, ….). We here focus on real-time estimation for static and dynamic models (a short introduction will be provided for statistic model). We provide a unified point of view for several parametric estimation problems which consist in finding a “good” estimation of Θ from an observed signal y = F (x, Θ) + n, where x is “true” signal and Θ is the parameters to be estimated and n is a noise corrupting the observation. In automatic control such class of related problems are: identifiability and identification of uncertain parameters in the system equations, including delays, (linear or non linear and even for closed loop systems); estimation of state variables, which are not measured (even for closed loop systems); fault diagnosis and isolation; observer-based chaotic synchronization, with applications in cryptography and secure systems. In signal, image and video processing such class of related problems are (noise removal): signal modelling, demodulation, restoration, (blind) equalization ; Data compression, Decoding for error correcting codes ; Detection of abrupt change, … Of course, such technics has deep impact on many applicative fields such as transportation, robotics, life sciences, … Here, case studies will be taken from robotics and life sciences: 1. localization of a Mobile robot (introduction to SLAM Simultaneous Localization And Mapping), 2. Quadrotot (posture estimation), 3. Posture estimation using accelerometer for squatt exercise (Biomechanics), 4. Environmental monitoring using oysters as bio-sensors,
Objectifs pédagogiques
At the end of the course, the student will be able to: - understand, - analyze, - and develop solutions, for various estimation problems. These estimation problems concern: identifiability and identification of uncertain parameters in the system equations, including delays, (linear or nonlinear and even for closed loop systems); estimation of state variables, which are not measured (even for closed loop systems); fault diagnosis and isolation; observer-based chaotic synchronization, localizability of mobile robots (including drones, wheeled mobile robots, and underwater vehicles), estimation of time derivative for noisy signal (with some applications in signal processing).
Objectifs de développement durable
Modalités de contrôle de connaissance
Contrôle Continu
Commentaires: Students will be evaluated on the basis of a project (case study from the worlds of robotics, living systems...). The projects will be presented at the beginning of the elective during the introductory session
Ressources en ligne
Some have to be developped - Pdf for each master session and practical session, - Matlab/Simulink code (complete solution or partial one depending on the context), - external web links (using Wikipedia & <https://fr.mathworks.com> webinar & online solution and courses)
Pédagogie
Project and case study. The plan giving the learning sequence is given below: I Introduction II Linear/non linear regression (introduction to satistic model) II Linear Model Based Technics 1. Observability, Identifiability, localizability (Robotics), … 2. Geometric framework 3. Algebraic framework 4. Linear design (Kalman/Luenberger observers, full/uncomplete estimator…) 5. To work or not to work with a linearized system ? III Non Linear Model Based Technics 1. Introduction to non linear problems (Observability, Identifiability, localizability, …) 2. Geometric framework 3. Algebraic framework 4. Uniform observability & Local decomposition 5. Non linear estimator design (High gain, Homogeneous, Sliding Mode) IV Ultra-local Model Based Technics (or Model free technics) 1. Introduction 2. Algebraic Annihilators 3. Parameters estimation 4. Real-time Derivative estimation 5. State estimation
Séquencement / modalités d'apprentissage
| Nombre d'heures en CM (Cours Magistraux) : | 16 |
|---|---|
| Nombre d'heures en TD (Travaux Dirigés) : | 18 |
| Nombre d'heures en TP (Travaux Pratiques) : | 14 |
| Nombre d'heures en Séminaire : | 0 |
| Nombre d'heures en Demi-séminaire : | 0 |
| Nombre d'heures élèves en TEA (Travail En Autonomie) : | 48 |
| Nombre d'heures élèves en TNE (Travail Non Encadré) : | 0 |
| Nombre d'heures en CB (Contrôle Bloqué) : | 0 |
| Nombre d'heures élèves en PER (Travail PERsonnel) : | 0 |
| Nombre d'heures en Heures Projets : | 0 |
Pré-requis
Some basic linear control technics (linear state feedback), basic mathematics (linear algebra, basic algebra (such as ring, group) and basic analysis (differentiation, …) and some basic physics (electrical laws and mechanics)
Nombre maximum d'inscrits
64