Reinforcement learning for system identification and optimal experiment design.
Identifying the parameters of a mechatronic system means exciting it and watching how it responds. Which excitation you choose determines how much you learn, and classical optimal experiment design solves for that choice up front, from a model you do not yet trust. My research treats it instead as a sequential decision problem: an agent picks the next experiment based on what the previous ones revealed, and learns a policy that generalises across systems.
Publications
Reinforcement Learning for Optimal Experiment Design in Parameter Identification of Mechatronic Systems
Julian Langschwert, Georg Schaefer, Jakob Rehrl, Stefan Huber, Simon Hirländer
DEXA International Workshop on Optimisation of Industrial Production with AI Algorithms · Conference paper · June 2026
A current list is kept on ORCID.
Affiliations
- IDA Lab Salzburg — with Simon Hirländer
- ISIA project, Fachhochschule Salzburg