Unknown Nonlinear Dynamic

Research on unknown nonlinear dynamics focuses on developing control strategies and state estimation techniques for systems whose behavior is not fully understood. Current approaches leverage data-driven methods, including Gaussian processes, neural ordinary differential equations (NODEs), and physics-guided neural networks, often incorporating linear autoregressive models to improve efficiency and interpretability. These advancements enable robust control and accurate state observation in complex systems, with applications ranging from robotics and autonomous driving to more general control engineering problems. The ultimate goal is to create adaptable and reliable control systems that can function effectively even with incomplete knowledge of the underlying dynamics.