In my research I approach control problems in a bio-inspired way. I use model predictive control (MPC), an important family of advanced control algorithms, calculating the best control input based on prediction of future evolution of the system. MPC algorithms require a model of the system - I explore advantages of using models learnt from data, particularly of using recurrent neural networks for this task. I also work on a hardware-aware optimizer for nonlinear functions which our MPC implementation uses to find the optimal control strategy.

Before joining INI I obtained my master’s degree in physics in 2020 from ETH Zurich. Already at that time I was interested in feedback control - not only as a tool facilitating experiment design, but also as a procedure inducing new, otherwise unobservable phenomena.