Hanna Krasowski

I develop and integrate formal methods that guide machine learning to obtain effective, data-efficient, safe, and interpretable models. This hybrid solution approach enables automation of real-world tasks, especially when high-quality data is expensive to obtain or available knowledge is abstract, like natural-language traffic rules and implicit system dynamics. I design and test algorithms for a variety of real-world systems, with a focus on maritime vessels and, more recently, systems biology.

Currently, I am a postdoctoral researcher in Murat Arcak’s group at UC Berkeley EECS and affiliated with Berkeley Artificial Intelligence Research (BAIR). I am co-advised by Sanjit Seshia. Previously, I was a doctoral researcher at Technical University of Munich’s Cyber-Physical Systems Group led by Matthias Althoff. In 2022, I visited Aaron Ames’ group at Caltech. I obtained a Master’s degree in Robotics, Cognition, Intelligence from Technical University of Munich and a Bachelor’s degree from Technical University of Darmstadt.

Research interests

machine learning \(\cdot\) formal methods \(\cdot\) autonomous systems \(\cdot\) systems biology

Software packages

CommonOcean is a collection of benchmarks for motion planning of autonomous vessels and provides researchers with software tools to build, evaluate, and compare their motion planners. CommonOcean includes a temporal logic formalization of maritime traffic rules, converters for AIS data, and vessel models with parameters for multiple vessel types.

CommonRoad-RL offers a reinforcement learning gym to solve motion planning problems from CommonRoad driving scenarios. Its modularity facilitates the benchmarking of autonomous driving research with reinforcement learning.

Beyond research

I am passionate about encouraging girls to consider STEM subjects for their careers. In my opinion, STEM subjects allow you to gain a deep understanding of the world and teach you how to manipulate it to achieve innovation. For me, these opportunities to shape the world are very empowering, and I hope to convey this motivation and spark interest in STEM by running workshops during which female role models guide high-school girls through hands-on science experiments. I head the initiative Girls macht MI(N)T! and actively support TUM Entdeckerinnen.