Workshop on Gaussian Process Learning for Systems and Control
IFAC World Congress 2023, Yokohama, Japan

The workshop on Gaussian Process Learning for Systems and Control will be held at IFAC World Congress 2023 on Sunday, 9th July 2023 with contributions by Thomas Beckers, Sandra Hirche, Rolf Findeisen, Maik Pfefferkorn, Colin Jones, and Karl Berntorp. Find the schedule and information about the speakers below.

One challenge in controller design is to achieve a desired performance and guarantee safe operation, e.g., via the satisfaction of constraints. This goal should be fulfilled even in the presence of uncertainties. Uncertainties can stem from unknown system dynamics, external disturbances, or interaction of the control system with unknown entities such as other controlled systems or humans. One way to deal with uncertainties is obtaining an estimate of them via machine learning techniques, such as Gaussian processes.

Gaussian processes have been used increasingly as a data-driven technique within the past two decades due to many beneficial properties such as the bias-variance trade-off and the close relation to Bayesian mathematics. In contrast to most of the methods, Gaussian process models provide a regression function and a measure for the uncertainty of the prediction. This powerful property makes them very attractive for many applications in control, e.g., model predictive control, robust control, reinforcement learning, and general optimization tasks, as the uncertainty measure allows providing convergence, performance, and safety guarantees. However, fusing/embedding machine learning and especially Gaussian processes in a closed-loop control system pose several challenges, such as closed-loop uncertainty propagation or real-time feasible online learning.

This tutorial-style workshop aims to provide insight into the fundamentals behind Gaussian processes for modeling and control and sketches some of the open challenges and opportunities using Gaussian processes for modeling and control. Experts/lecturers with experience in Gaussian processes and (optimization-based) control from academia and industry will introduce Gaussian processes’ basics and spotlight Gaussian processes’ opportunities for the control community and recent advances in learning-based control under uncertainties in general. The workshop targets an audience from graduate level to experienced theoretical and practically oriented control engineers who aim to improve their knowledge in controller design under uncertainties leveraging Gaussian processes and machine learning. It was run in a similar structure first at the 61st IEEE Conference on Decision and Control in 2022 in Cancun, where it attracted about 40 participants. For the IFAC World Congress we expect a similar, or even higher number of participants.

Time Type Content Speaker
08:15 – 08:30 Warming up: Welcome, motivation, and introduction
08:30 – 09:15 Tutorial From Gaussian distributions to Gaussian processes Thomas Beckers
09:15 – 10:00 Tutorial Regression with Gaussian processes Maik Pfefferkorn
Coffee break
10:30 – 11:30 Tutorial Learning dynamical systems with GPs Thomas Beckers
11:30 – 13:00 Tutorial Control of dynamical systems using GPs: MPC, safety and
guarantees 		Maik Pfefferkorn
Lunch break
14:00 – 14:45 Spotlight Industrial perspectives on Gaussian process learning-based
control 		Karl Berntorp
14:45 – 15:30 Spotlight Efficient online learning in closed loop control of dynamical
systems 		Sandra Hirche
Coffee break
16:00 – 16:45 Spotlight Exploiting Gaussian Processes in Predictive Control Rolf Findeisen
16:45 – 17:30 Spotlight Learning, optimization, and control with application to-
wards energy efficiency 		Colin Jones
17:30 – 18:00 Closing remarks/discussion

Thomas Beckers

is an incoming Assistant Professor at Vanderbilt University. Currently, he is a postdoctoral researcher at the Department of Electrical and Systems Engineering, University of Pennsylvania. He is member of the GRASP Lab and the PRECISE Center. In 2020, he earned his doctorate in Electrical Engineering at the Technical University of Munich (TUM), Germany. He received the B.Sc. and M.Sc. degree in Electrical Engineering in 2010 and 2013, respectively, from the Technical University of Braunschweig, Germany. In 2018, he was a visiting researcher at the University of California, Berkeley. He is a DAAD AInet fellow and was awarded with the Rhode & Schwarz Outstanding Dissertation price. His research interests include physics-enhanced learning, nonparametric models, and safe learning-based control.

Sandra Hirche

received the Diplom-Ingenieur degree in aeronautical engineering from Technical University Berlin, Germany, in 2002 and the Doktor-Ingenieur degree in electrical engineering from Technical University Munich, Germany, in 2005. From 2005 to 2007 she was awarded a Postdoc scholarship from the Japanese Society for the Promotion of Science at the Fujita Laboratory, Tokyo Institute of Technology, Tokyo, Japan. From 2008 to 2012 she has been an associate professor at Technical University Munich. Since 2013 she is TUM Liesel Beckmann Distinguished Professor and has the Chair of Information-oriented Control in the Department of Electrical and Computer Engineering at Technical University Munich. Her main research interests include cooperative and distributed networked control as well as learning control with applications in human–robot interaction, multi-robot systems, and general robotics. She has published more than 150 papers in international journals, books, and refereed conferences.

Rolf Findeisen

studied engineering cybernetics at the University of Stuttgart and chemical engineering at the University of Wisconsin – Madison. He began his doctoral studies at ETH Zurich’s, which he completed in 2004 following his doctoral father to the University of Stuttgart. 2007, Rolf was appointed professor at the Institute of Automatic Control at Otto-von-Guericke University Magdeburg. Since August 2021, he heads the Control and Cyber-physical Systems Laboratory at the Technical University of Darmstadt. Rolf is engaged in method development in the area of systems theory and control engineering, focusing on optimization-based and predictive control; fusing machine learning approaches such as Gaussian processes and neural networks with model based control providing guarantees; and control of complex, distributed systems via communication networks.

Maik Pfefferkorn

obtained his Bachelor’s and Master’s degrees in biosystems engineering from Otto-von-Guericke University Magdeburg in 2018 and 2020, respectively. Since 2020, he is a Ph.D. student in Rolf Findeisen’s group at the Otto-von-Guericke University Magdeburg. Maik’s research interests are in the theory and application of machine learning approaches, especially Gaussian process regression, in model predictive control with stability and safety guarantees.

Karl Berntorp

is Senior Principal Research Scientist in the Mitsubishi Electric Research Laboratories (MERL). He received his Ph.D. in 2014 at the Department of Automatic Control, Lund University, Sweden. Karl’s research interests include statistical signal processing, Bayesian inference and learning, sensor fusion, and optimization-based control, with applications to automotive, transportation, navigation, positioning, and communication systems. Karl is an Associate Editor and member of the IEEE Technology Conferences Editorial Board.

Colin Jones

is an Associate Professor in the Automatic Control Laboratory at the Ecole Polytechnique Federale de Lausanne (EPFL) in Switzerland. He was a Senior Researcher at the Automatic Control Lab at ETH Zurich until 2011 and obtained a PhD in 2005 from the University of Cambridge for his work on polyhedral computational methods for constrained control. Prior to that, he was at the University of British Columbia in Canada, where he took a BASc and MASc in Electrical Engineering and Mathematics. Colin has worked in a variety of industrial roles, ranging from commercial building control to the development of custom optimization tools focusing on retail human resource scheduling. His current research interests are in the theory and computation of predictive control and optimization, and their application to green energy generation, distribution and management.