Generative AI for Control Systems: Learning to Design Stable Controllers
Matteo Cercola at ECE 2026 Spring Colloquium
The rise of Generative Artificial Intelligence is fundamentally reshaping the engineering landscape. While originally developed for natural language processing and image synthesis, recent research trends demonstrate that these models possess vast potential far beyond creative content generation.
This lecture explores the emerging synergy between generative modeling and control theory. Specifically, I will present a novel framework for the synthesis of linear controllers by reframing the design process as a generative task. By leveraging Diffusion models, we bridge the gap between high-dimensional generative architectures and the rigorous requirements of system stability, offering a transformative approach to controller synthesis.