Project description
The project investigates the problem of characterizing uncertainty in control design approaches where controllers are synthesised using noisy data. In indirect data-driven control first a mathematical model of the system is identified from data and then classical model-based control strategies are employed for design; direct data-driven control methods unify these two steps by building maps from trajectories to controllers. The project aims at studying the fundamental uncertainty propagation mechanisms arising in these two approaches by leveraging mathematical tools from statistical learning theory, information theory, and system identification. The overarching goal is to determine whether, and in which scenarios, one approach is preferable to the other when robustness guarantees must be provided to guarantee safety.
Project information
Project Name | Uncertainty quantification for indirect and direct data-driven control |
Project leaders | Andrea Iannelli (Peter Eberhard) |
Project staff | Nicolas Chatzikiriakos |
Project duration | January 2023 - December 2025 |
Project number | PN 4-10 |
Publications PN 4-10
2024
- B. Song and A. Iannelli, “The role of identification in data-driven policy iteration: A system theoretic study,” International Journal of Robust and Nonlinear Control, vol. n/a, Art. no. n/a, 2024, doi: https://doi.org/10.1002/rnc.7475.