17/12/2019 - 10H00
Séminaire CID

 

Paulo Tabuada (professeur à UCLA)
Dragan Nesic (professeur à l'université de Melbourne)


Paulo Tabuada (professeur à UCLA) et Dragan Nesic (professeur à l'université de Melbourne) donneront chacun un séminaire le mardi 17 décembre prochain dans la salle 124 jaune du CRAN à l'ENSEM à partir de 10h.

 

Paulo Tabuada

One ounce of modeling is worth a pound of training: Data-driven control for nonlinear systems

Current learning-based techniques for the control of physical systems, such as reinforcement learning, require the crunching of large amounts of data for extended periods of time. In this talk we show how to obviate this hunger for data by judicious modeling. In particular, we will show how to control unknown nonlinear systems without prior data or training. Key to our approach is the re-interpretation of several results in control theory, such as Fliess and co-workers intelligent-PIDs, feedback linearization, and adaptive control, as different examples of data-driven control. We illustrate the usefulness and applicability of the results via experimental results and conclude by speculating about the right mix of model-based and data-driven design in the context of autonomous cyber-physical systems.

Dragan Nesic

Estimation framework for nonlinear systems with time scale separation

The estimation of unmeasured variables is a central objective in a broad range of applications. This problem turns into a challenging task when the underlying model is nonlinear and even more so when additionally it exhibits multiple time scales. This research work focuses on generating an estimation framework for a nonlinear systems exhibiting time scale separation. First, we overview observer design for slow states of general nonlinear singularly perturbed plants. Then, we address the full state estimation problem by introducing a multi-observer approach or nonlinear systems with slowly varying-time parameters. Moreover, we propose a new estimation technique that leads to a switched observer for the estimation of slow and fast state.