MS 4: Data-Integrated Control with Guarantees

This minisymposium is devoted to recent developments in designing data-integrated control systems while being able to provide rigorous guarantees for the overall system behavior. Research in recent years has shown surprising successes with data-based controllers reaching remarkable practical performance even without having an explicit model of the controlled system. However, providing formal guarantees regarding stability and robustness has remained a fundamental challenge. In particular, existing data-based approaches often come close to but cannot reliably exceed the performance of expert-designed controllers, especially in data-scarce settings. The research presented in this symposium goes beyond that by striving for provable guarantees for data-integrated control systems, e.g., by fusing model-based paradigms with data-driven methods. The results presented in the talks will lay the rigorous groundwork for data-integrated control in modern applications such as autonomous driving, power generation, or human-robot interaction, which are of great societal significance and demand reliable operation. The minisymposium will foster and stimulate scientific exchange and discussions between data scientists, system theorists, control engineers, and experts in first-principles modeling.