Luenberger Observer

Luenberger observers are linear state estimators used to reconstruct the internal state of a system from its output measurements. Current research focuses on extending their application to nonlinear systems, often employing neural networks, particularly neural ordinary differential equations (NODEs), to learn the observer's parameters or to facilitate transformations to more amenable coordinate systems. This work addresses challenges in handling system uncertainties and improving robustness, with applications ranging from robotic control using visual feedback to super-resolution of turbulent flows by incorporating physics-informed constraints within the observer design. These advancements enhance the applicability of Luenberger observers to complex, real-world systems where accurate state estimation is crucial.