PhD student Mr. Rajpal Singh

ME – PhD Thesis Colloquium Continual Koopman Learning for Data-Driven Control of Nonlinear Systems

A physics-informed Koopman representation based on generalized momenta is introduced, yielding a linear control-affine model in lifted coordinates with known input structure that avoids the bilinear state – input coupling inherent in standard Koopman approaches, enabling improved prediction accuracy and tractable controller synthesis.