Title: Learning Lipschitz Feedback Policies from Expert Demonstrations: Closed-Loop Guarantees, Robustness and Generalization Authors: Abed AlRahman Al Makdah; Vishaal Krishnan; Fabio Pasqualetti Date of Publication: 17 June 2022 Link: ieeexplore.ieee.org/document/97988…

Learning LQG controllers is easy when you have access to an optimal control sequence, and easier when you have seen a number of tasks. No need to know the system, the cost function, and the noise statistics! arxiv.org/abs/2303.09002 Taosha Guo Abed AlRahman Al Makdah @VishaalKrishna1

Data-driven control for network and multi-agent systems, via finite-time distributed computation and with robustness guarantees 👇🏽 ieeexplore.ieee.org/document/10076… Federico Celi

The LQG controller can be learned from offline open-loop data without knowing the system model and noise statistics. Here's how, together with error bounds and explicit data-driven formulas for the LQR/LQG gains and Kalman filter: arxiv.org/abs/2304.00381 Abed AlRahman Al Makdah