Hanna Krasowski, M.Sc.

Hanna Krasowski

Technische Universität München

Informatik 6 - Professur für Cyber Physical Systems (Prof. Althoff)

Postadresse

Postal:
Boltzmannstr. 3
85748 Garching b. München

Tel.: +49 (89) 289 - 18131
Sprechstunde: Thursday 1pm - 2pm
Raum: 5607.03.061
Homepage: https://hanna.krasowski.io
hanna.krasowski@tum.de

Curriculum Vitae

Hanna Krasowski joined the Cyber Physical Systems Group as a PhD candidate under the supervision of Prof. Dr.-Ing. Matthias Althoff in 2020. She is a member of the DFG Research Training Group on Continuous Verification of Cyber-Physical Systems (ConVeY). She visited the research group of Prof. Aaron Ames at the California Institute of Technology from July to December 2022. Hanna received her master's degree in Robotics, Cognition, Intelligence from Technical University of Munich in 2020 and her bachelor's degree in Mechanical and Process Engineering from Technical University of Darmstadt in 2017.

Her research interests include safe reinforcement learning, motion planning and formal methods. You can find more information on her personal webpage.

Offered Thesis Topics

I am always looking for self-motivated students to solve interesting problems arising in my research areas. If you are interested in my research and want to write a thesis in this field, simply send me a mail briefly describing your motivation.

Currently Available

—

Ongoing

[MT] How to Achieve Optimality in Safe Reinforcement Learning? (co-advised with Hannah Markgraf and Lukas Schäfer)

Finished

[MT] Long-term Horizon Planning for Underactuated Autonomous Vessels (co-advised with Marius Wiggert, UC Berkeley) - submitted May 2023
[MT] Safe Motion Planning for Underactuated Autonomous Vessels (co-advised with Marius Wiggert, UC Berkeley) - submitted May 2023
[BT] Traffic Rule Compliant Simulation Environment for Marine Motion Planning - submitted April 2023
[MT] Motion Planning for Sustainable Autonomous Vessel (co-advised with CargoKite) - submitted January 2023
[MT] Generalizing Marine Traffic Rules for Multiple Vessel Types and Emergencies - submitted August 2022
[MT] Generating Near-collision Scenarios from Marine Traffic Data - submitted May 2022
[BT] Control Benchmarking of Ship Models (co-advised with Victor Gaßmann) - submitted March 2022
[BT] Reachset Conformance for Dynamic Vessel Models - submitted February 2022
[MT] Multi-agent Reinforcement Learning for Autonomous Vessels - submitted January 2022
[BT] Benchmarking Provably Safe Reinforcement Learning Approaches (co-advised with Xiao Wang) - submitted October 2021
[BT] Simulation Environment for Marine Motion Planning - submitted September 2021
[MT] Safe and Efficient Reinforcement Learning for Autonomous Driving in Urban Scenarios - submitted April 2021
[BT] Generation of Benchmarks for Marine Motion Planning - submitted April 2021

Teaching

Lectures

Formal Methods for Cyber-Physical Systems [WiSe 20/21, WiSe 21/22] – Safe Reinforcement Learning
Cyber-Physical Systems [SoSe 21, SoSe 22] – Discrete Systems

Practical Course – Motion Planning for Autonomous Vehicles [WiSe 20/21, SoSe 21, WiSe 21/22, SoSe 22, SoSe 23]

Benchmarking Marine Motion Planning
Reinforcement Learning for Autonomous Vessels
Set-based Prediction of Vessels
Developing an Autonomous Vessel Simulation
Motion Planning for Autonomous Vessels

Seminar – Cyber-Physical Systems [WiSe 20/21, SoSe 21, WiSe 21/22, SoSe 22]

Review on Motion Planning and Control Strategies for Autonomous Vessels
Safe Reinforcement Learning for Motion Planning
Dynamic Vessel Models and their Applications
Safe Reinforcement Learning with Logical Specifications
Safe Multi-Agent Reinforcement Learning

Tool CommonOcean

CommonOcean is a collection of composable benchmarks for motion planning of autonomous vessels and provides researchers with means of evaluating and comparing their motion planners. A benchmark consists of a scenario with a planning problem, a vessel model including vessel parameters, and a cost function composing a unique id. Along with benchmarks, we provide tools for motion planning.

Publications

2023

Niklas Kochdumper; Hanna Krasowski; Xiao Wang; Stanley Bak; Matthias Althoff: Provably Safe Reinforcement Learning via Action Projection using Reachability Analysis and Polynomial Zonotopes. IEEE Open Journal of Control Systems, 2023 mehr… BibTeX Volltext ( DOI )

2022

Hanna Krasowski; Matthias Althoff: CommonOcean: Composable Benchmarks for Motion Planning on Oceans. 2022 IEEE International Conference on Intelligent Transportation Systems (ITSC) , 2022 mehr… BibTeX Volltext ( DOI ) Volltext (mediaTUM)
Hanna Krasowski; Yinqiang Zhang; Matthias Althoff: Safe Reinforcement Learning for Urban Driving using Invariably Safe Braking Sets. 2022 IEEE International Conference on Intelligent Transportation Systems (ITSC) , 2022 mehr… BibTeX Volltext ( DOI ) Volltext (mediaTUM)

2021

Hanna Krasowski; Matthias Althoff: Temporal Logic Formalization of Marine Traffic Rules. 2021 IEEE Intelligent Vehicles Symposium (IV), 2021 mehr… BibTeX Volltext ( DOI ) Volltext (mediaTUM)
Xiao Wang; Hanna Krasowski; Matthias Althoff: CommonRoad-RL: A Configurable Reinforcement Learning Environment for Motion Planning of Autonomous Vehicles. IEEE International Conference on Intelligent Transportation Systems (ITSC), 2021 mehr… BibTeX Volltext ( DOI ) Volltext (mediaTUM)

2020

Hanna Krasowski; Xiao Wang; Matthias Althoff: Safe Reinforcement Learning for Autonomous Lane Changing Using Set-Based Prediction. 2020 IEEE International Conference on Intelligent Transportation Systems (ITSC), 2020 mehr… BibTeX Volltext ( DOI ) Volltext (mediaTUM)