Laura Boca de Giuli, a PhD student in Information Technology at the Department of Electronics, Information and Bioengineering (DEIB) at Politecnico di Milano, received the Honorable Mention as finalist for the Young Author Theory Paper Prize at the IFAC World Congress 2026, for the paper “Learning-based predictive control with Bayesian neural networks under safety guarantees”. The Congress took place in Busan, South Korea, from August 23 to 28, 2026, and brought together more than 4000 participants from around the world.
The paper was co-authored with Alessio La Bella from DEIB and Manish Prajapat, Johannes Köhler, Anna Scampicchio, and Melanie Zeilinger from the Intelligent Control Systems Group at ETH Zurich, where Laura spent five months as a visiting researcher in 2025. The work proposes a novel control algorithm that leverages Bayesian neural networks to learn online an unknown system, while maintaining high computational efficiency and guaranteeing the satisfaction of safety constraints throughout the learning process. A distinctive feature of the approach is its ability to safely explore the system dynamics to collect informative data, while guaranteeing that the exploration phase terminates in a finite number of steps.
The algorithm was validated in simulation on a district heating network and has recently been implemented on a real-world plant, showing strong control performance and confirming the potential of the proposed approach for complex energy applications.