Nov.26, 2022｜11:00-12:00｜Room 316-1, Building 133
Prof. Rolf Findeisen (Univ of Magdeburg, Germany)
Abstract/Biography
Many autonomous systems, such as robots, drones, or autonomous cars, should operate safely yet optimally under uncertain conditions. Often the model of the environment or the system itself is unknown or only partially known, and uncertainties and disturbances are present. We outline that fusing predictive control with machine learning approaches, such as Gaussian processes or Bayesian Neural Networks, allows for safe, optimal control and planning for autonomous systems.
Learning allows adapting based on measurements and data and individualizing the control and planning task to the individual job, while predictive control ensures safety and optimality by exploiting the learned model, cost function, reference, or disturbance. We outline how to integrate systems knowledge and properties, such as controllability and constraint satisfaction, into learning to achieve safe, stable, and optimal behavior. The presented methods are underlined, considering the control and planning spanning from the operation of robots, drones, and autonomous driving to quantum dot microscopy.Rolf Findeisen studied engineering cybernetics at the University of Stuttgart and chemical engineering at University of Wisconsin-Madison. After his studies, he joined the Institut for Automatic Control at ETH Zürich, before moving with his advisor (Frank Allgöwer) to the University of Stuttgart, where he obtained his Dr.-Ing. In 2007 he joined the Otto-von-Guericke University Magdeburg where he heads the Laboratory for Systems Theory and Automatic Control. He has spent time as a visiting professor at the Massachusetts Institute of Technology, EPF Lausanne, Imperial College London and the Mitsubishi Research Laboratory in Cambridge. Rolf is editor and associated editor of several journals such as the IEEE Transactions on Control of Network Systems and the IEEE Control Systems Magazine. He is actively involved in the organization of international conferences and workshops, such as the World Congress of the International Federation of Automatic Control (IFAC) 2020 in Berlin or the conference on the Foundations of Systems Biology in Engineering (FOSBE) 2016.