Leo Simpson

External PhD Student

Short Bio

Leo Simpson was born Toulouse, France, in 1996. After studying scientific topics during two years of “Classe Préparatoire”, he studied at Ecole Polytechnique from 2016 to 2019, for a master of applied mathematics. During these studies, he has been focusing on mathematics, and more, in a multidisciplinary formation involving physics, mechanics, chemistry, and engineering sciences. 

He pursued his studies doing a master of mathematics at the Technical University of Munich from 2019 to 2021. This master’s degree was focused on optimization, and his master thesis topic was numerical optimal control applied to walking robots, in a group of researchers from Siemens Technology. He then decided to start his Ph.D. in Tool-temp, under the supervision of Dr. Jonas Asprion and Prof. Moritz Diehl, in the framework of the Marie Curie Initial Training Network “ELO-X”.


Project Description at Tool-Temp AG

Tool-temp is a company producing Temperature Control Units, tools that control the temperature of a variety of industrial processes by means of circulating a thermal fluid. The dynamics of temperature transfers involved in these processes can be quite complex and are hard to model as they strongly depend on unknown parameters of the industrial process. The aim of this Ph.D. project is to investigate algorithms that simultaneously learn the parameters of these dynamics and perform a suitable control policy to control the temperature over these dynamics. 

Both numerical and real machine experiments will be considered, to assess the quality of different algorithms. Machine learning techniques and control theory will be the most relevant mathematical concepts in this research project.


Recent pre-prints

Léo Simpson, Armin Nurkanović, Moritz Diehl. Direct Collocation for Numerical Optimal Control of Second-Order ODE, 2022.

Léo Simpson, Andrea Ghezzi, Jonas Asprion, Moritz Diehl. An Efficient Method for the Joint Estimation of System and Noise Covariances for Linear Time-Variant Systems, 2023

Jing Xie, Léo Simpson, Jonas Asprion, Riccardo Scattolini. A Learning-based Model Predictive Control Scheme with Application to Temperature Control Units, 2024.