Unknown Linear System

Research on unknown linear systems focuses on developing data-driven methods to learn and control systems whose dynamics are initially uncharacterized. Current efforts concentrate on improving the efficiency and robustness of algorithms, addressing challenges like noise, adversarial perturbations, and the curse of dimensionality in learning stabilizing controllers. This work leverages techniques such as Gaussian processes, set membership estimation, and online learning algorithms to achieve better performance guarantees and reduced computational complexity. The advancements in this field have significant implications for robust control design and various applications requiring adaptive control in uncertain environments.