Probabilistic Control

During my studies of Engineering Cybernetics I wondered why it is so hard to reach high control performance on real systems. It seemed so easy in the simulations! The problem is that the real world is never perfect: Imprecisions in the mechanics, noise in the sensors and model errors make it hard to build good controllers.

Therefore, my focus lies on the probabilistic view on controllers as well as the connections between control theory and machine learning.

I am co-affiliated with ETH Zurich, where I am involved in the MPC with Gaussian Processes project.

After finishing my PhD at ETH Zurich, I moved to the Bosch Center for Artificial Intelligence.