Data-based modeling and control

We develop analysis, modeling and synthesis tools for closed-loop systems featuring data-driven (or learning) components, with the goal of better understanding their fundamental limitations and robustness properties. Based on this we propose improved solutions that provide theoretical guarantees together with computationally attractive procedures.

We consider both indirect (or system identification-based) approaches, where first a model is identified and then used for control design; and direct approaches, where data are used for control design. A special focus is on explicitly accounting for uncertainty arising from noisy and finite data. To this end, we study algorithms to (i) efficiently generate informative data, (ii) identify unknown dynamical systems from such data, and (iii) construct models and uncertainty estimates that enable end-to-end guarantees in control.

We recognize the interplay between data collection, system identification, and (robust) controller design as key to understand the fundamental limits of learning dynamical systems and how these limits impact applicability of data-driven methods and their achievable performance.

Our research combines tools from high-dimensional statistics, information theory, theoretical machine learning, and traditional topics in control and systems theory (system identification, nonlinear and robust control) to derive rigorous guarantees for learning from data trajectories in dynamic, uncertain and stochastic environments.

System identification

• Chatzikiriakos N., Jamieson, K., Iannelli, A. - "High Effort, Low Gain: Fundamental Limits of Active Learning for Linear Dynamical Systems" - 29. International Conference on Artificial Intelligence and Statistics (spotlight), 2026. Preprint Link
• Chatzikiriakos N., Strässer R., Allgöwer F., Iannelli A. - "End-to-end guarantees for indirect data-driven control of bilinear system with finite stochastic data" - Automatica, 2026. Link 
• Chatzikiriakos N., Iannelli A. - "Sample Complexity Bounds for Linear System Identification from a Finite Set" - IEEE Control Systems Letters 2024. Link
• Ozan D.E., Yin M., Iannelli A., Smith R.S. - "Kernel-Based Identification of Local Limit Cycle Dynamics with Linear Periodically Parameter-Varying Models" - IEEE Conference on Decision and Control 2022. Link
• Iannelli A., Fasel U., Smith R.S. - "The Balanced Mode Decomposition Algorithm for Data-Driven LPV Low-Order Models of Aeroservoelastic Systems" - Aerospace Science and Technology, 2021. Link
• Yin M., Iannelli A., Smith R.S. - "Subspace Identification of Linear Time-Periodic Systems with Periodic Inputs" - IEEE Control Systems Letters, 2021. Link

Data-driven control

• Bosso A., Borghesi M., Iannelli A., Notarstefano G., Teel A.R. - "Data-Driven Control of Continuous-Time LTI Systems via Non-Minimal Realizations" - IEEE Transactions on Automatic Control, 2026. Link
• Iannelli A., Postoyan R. - "A hybrid systems framework for data-based adaptive control of linear time-varying systems - IEEE Transactions on Automatic Control, 2025. Link
• Yin M., Iannelli A., Smith R.S. - "Maximum Likelihood Estimation in Data-Driven Modeling and Control" - IEEE Transactions on Automatic Control, 2021. Link

K. Jamieson (University of Washington), M. Yin (Leibniz University Hannover), R. S. Smith (ETH Zürich), A. Bosso (University of Bologna), R. Postoyan (CNRS, Université de Lorraine).