KKL Observer Synthesis for Nonlinear Systems via Physics-Informed Learning

M. Umar B. Niazi, John Cao, Matthieu Barreau, Karl H. Johansson · 2025

Read KKL Observer Synthesis for Nonlinear Systems via Physics-Informed Learning by M. Umar B. Niazi, John Cao, Matthieu Barreau, Karl H. Johansson (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-11655.txt.

Description

Umar Niazi and three co-authors tackle a problem in control engineering: estimating the hidden state of a nonlinear system from its outputs. One elegant design, the KKL observer, maps the state into a higher-dimensional space where the dynamics become simple and stable, then maps the estimate back. Finding those two maps is the hard part, and it has limited the method's use.

The authors teach the forward map to a neural network that is trained to respect the system's physics, and the return map to a plain network. They prove bounds showing how estimation error depends on how well the networks learn and on noise in the system. Tests on standard example systems show the learned observer keeps working outside the range it was trained on, better than current methods. The work aims to make this observer design practical for real systems.

Catalog record

CREATORM. Umar B. Niazi, John Cao, Matthieu Barreau, Karl H. Johansson
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE79 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-11655

Browse the full library · About the Tower of Babel