I am a PhD student in the Embodied Vision group. My research interest lies at the intersection of Machine Learning and Control. I find it particularly interesting to model and control systems based on high-dimensional observations (e.g. images or videos) employing probabilistic dynamics modeling in a data-efficient manner. Of further interest is the physical interpretability of learnt representations.
Before starting in the Embodied Vision group, I studied electrical engineering at RWTH Aachen, Germany and wrote my master's thesis about probabilistic compression and uncertainty in deep neural networks in cooperation with the Bosch Center for Artificial Intelligence, Renningen.
Proceedings of the 2nd Conference on Learning for Dynamics and Control (L4DC), 120, pages: 640-650, Proceedings of Machine Learning Research (PMLR), (Editors: Alexandre M. Bayen and Ali Jadbabaie and George Pappas and Pablo A. Parrilo and Benjamin Recht and Claire Tomlin and Melanie Zeilinger), 2020, arXiv:2005.03770 (conference)
In Conference on Robot Learning 2020, 2020 (inproceedings)
Trajectory optimizers for model-based reinforcement learning, such as the Cross-Entropy Method (CEM), can yield compelling results even in high-dimensional control tasks and sparse-reward environments. However, their sampling inefficiency prevents them from being used for real-time planning and control. We propose an improved version of the CEM algorithm for fast planning, with novel additions including temporally-correlated actions and memory, requiring 2.7-22x less samples and yielding a performance increase of 1.2-10x in high-dimensional control problems.
Our goal is to understand the principles of Perception, Action and Learning in autonomous systems that successfully interact with complex environments and to use this understanding to design future systems