Seminar: "Data-Enabled Predictive Control (DeePC)"

As part of the 2025 DisCoRaP seed project, GiPSA-Lab will host Florian Dörfler, a professor at ETH Zurich, on March 36 for a seminar on modern data-driven control approaches.

During this seminar, Florian Dörfler will present an innovative approach to predictive control that enables the optimization and constraint of system control without an explicit model, directly from experimental data. This method, known as hashtag#DeePC, relies on a behavioral representation of systems and demonstrates, in the deterministic linear case, equivalence with Model Predictive Control (MPC) while often offering better performance in practice.

The seminar will also cover:
• robustness against noise and nonlinearities
• extensions to stochastic systems via a Gaussian approach
• applications in aerial robotics, power electronics, and electrical systems