Abstract: Data availability and advances in robotics are driving forces for innovations in many fields (manufacturing, energy systems, transportation, …). They require adaptive control and optimization methods for enhanced functionality and new economic models. The distinct challenges of advanced manufacturing - process variability, drifts, and lack of in-situ measurements - demand approaches that are data-efficient, conform to physical process limits, and prioritize safety.
In this talk I will discuss data-driven, predictive, and learning-based optimization methods which enhance productivity, quality, and energy consumption, while respecting safety and operational constraints and making the most out of the available data. I will highlight Bayesian optimization algorithms for optics alignment and process control, approaches for repetitive control incorporating data from the system to accomodate for process changes and drifts, and integrating digital twins in such algorithms to increase productivity on a machine level, as well as on a system level, while making optimal use of the available resources. These algorithms are inspired by real-world applications including additive manufacturing, plasma spray coating, and high-precision motion stages, and I will showcase their application on such systems.
Short bio: Alisa Rupenyan holds the endowed professorship in Industrial AI from the Johann Jakob Rieter foundation at the Zurich University for Applied Sciences and specializes in continuous optimization and automation of industrial systems. Until September 2023 she was a PI and senior scientist at the Automatic Control Laboratory at ETH Zurich and at the same time group leader for Automation at Inspire, the technology transfer institute at ETH Zurich.
She has a PhD in Physics from Vrije University Amsterdam, and MSc and BSc degrees in Engineering Physics from Sofia University. She was granted an ETH excellence postdoctoral fellowship to fund her research in the field of high-harmonic spectroscopy for the study attosecond molecular dynamics. After that, she switched to the industry and lead a team in a Swiss robotics startup combining her experience in spectroscopy with machine learning and control. Her research interests include autonomous machines, decision-making in industrial settings, learning-based optimization and control. She is an active member of the National Center for Competence in Research (NCCR) Automationfunded by SNSFin Switzerland, leading the efforts in automation for advanced manufacturingand related applications.
Go to the Events page.