Learning Robust State Observers using Neural ODEs

Keyan Miao, Konstantinos Gatsis
Proceedings of The 5th Annual Learning for Dynamics and Control Conference, PMLR 211:208-219, 2023.

Abstract

Relying on recent research results on Neural ODEs, this paper presents a methodology for the design of state observers for nonlinear systems based on Neural ODEs, learning Luenberger-like observers and their nonlinear extension (Kazantzis-Kravaris-Luenberger (KKL) observers) for systems with partially-known nonlinear dynamics and fully unknown nonlinear dynamics, respectively. In particular, for tuneable KKL observers, the relationship between the design of the observer and its trade-off between convergence speed and robustness is analysed and used as a basis for improving the robustness of the learning-based observer in training. We illustrate the advantages of this approach in numerical simulations.

Cite this Paper

BibTeX
@InProceedings{pmlr-v211-miao23a, title = {Learning Robust State Observers using Neural ODEs}, author = {Miao, Keyan and Gatsis, Konstantinos}, booktitle = {Proceedings of The 5th Annual Learning for Dynamics and Control Conference}, pages = {208--219}, year = {2023}, editor = {Matni, Nikolai and Morari, Manfred and Pappas, George J.}, volume = {211}, series = {Proceedings of Machine Learning Research}, month = {15--16 Jun}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v211/miao23a/miao23a.pdf}, url = {https://proceedings.mlr.press/v211/miao23a.html}, abstract = {Relying on recent research results on Neural ODEs, this paper presents a methodology for the design of state observers for nonlinear systems based on Neural ODEs, learning Luenberger-like observers and their nonlinear extension (Kazantzis-Kravaris-Luenberger (KKL) observers) for systems with partially-known nonlinear dynamics and fully unknown nonlinear dynamics, respectively. In particular, for tuneable KKL observers, the relationship between the design of the observer and its trade-off between convergence speed and robustness is analysed and used as a basis for improving the robustness of the learning-based observer in training. We illustrate the advantages of this approach in numerical simulations.} }
Endnote
%0 Conference Paper %T Learning Robust State Observers using Neural ODEs %A Keyan Miao %A Konstantinos Gatsis %B Proceedings of The 5th Annual Learning for Dynamics and Control Conference %C Proceedings of Machine Learning Research %D 2023 %E Nikolai Matni %E Manfred Morari %E George J. Pappas %F pmlr-v211-miao23a %I PMLR %P 208--219 %U https://proceedings.mlr.press/v211/miao23a.html %V 211 %X Relying on recent research results on Neural ODEs, this paper presents a methodology for the design of state observers for nonlinear systems based on Neural ODEs, learning Luenberger-like observers and their nonlinear extension (Kazantzis-Kravaris-Luenberger (KKL) observers) for systems with partially-known nonlinear dynamics and fully unknown nonlinear dynamics, respectively. In particular, for tuneable KKL observers, the relationship between the design of the observer and its trade-off between convergence speed and robustness is analysed and used as a basis for improving the robustness of the learning-based observer in training. We illustrate the advantages of this approach in numerical simulations.
APA
Miao, K. & Gatsis, K.. (2023). Learning Robust State Observers using Neural ODEs. Proceedings of The 5th Annual Learning for Dynamics and Control Conference, in Proceedings of Machine Learning Research 211:208-219 Available from https://proceedings.mlr.press/v211/miao23a.html.

Related Material