Education

University of Tuebingen & MPI-IS
PhD candidate in Computer Science, advised by Georg Martius.
Nov 2023 -
KTH Stockholm
MSc in Computer Science, focused on Robotics and Deep Learning
Fall 2018 - Spring 2021
RWTH Aachen University
MSc in Electrical Engineering, Information Technology, and Computer Engineering, focused on Control Theory. Thesis advised by Michael Muehlebach and Sebastian Trimpe.
Fall 2018 - Spring 2021
RWTH Aachen University
B.S. in Electrical Engineering, Information Technology, and Computer Engineering, advised by Ehsan Zandi.
Fall 2015 - Spring 2018

Employment

AL, University of Tübingen & MPI-IS
PhD Candidate in the Autonomous Learning Group under Georg Martius.
Nov 2023 - Present
Tübingen, Germany
EV, MPI-IS
Researcher in the Embodied Vision Group. Worked on differentiable physics simulation for rigid bodies to perform simultaneous system identification and control for robot manipulators.
Aug 2021 - Feb 2022
Tübingen, Germany
LDS, MPI-IS
MSc. Thesis in the Learning and Dynamical Systems Group under Michael Muehlebach. Used Iterative Learning Control to teach robots how to juggle.
Aug 2021 - Feb 2022
Tübingen, Germany
i6, RWTH Aachen University
As a student researcher in the Human Language Technology and Pattern Recognition Group under Albert Zeyer, I worked on end-to-end automatic speech recognition systems using Transducer models. Focused on improving external LLM fusion and related training strategies.
Jan 2021 - Jul 2021
Aachen, Germany
IAV Automotive Engineering
As an intern in the Perception Group, designed and implemented a platform-independent sensor fusion library in C++.
Jul 2020 - Dec 2020
Dresden, Germany
WZL, RWTH Aachen University
At WZL, worked alongside an industry project to develop smart ticket barriers to prevent fare evasion. Contributed to driver development, real-time sensor visualization, sensor integration in AWS Greengrass, and porting human tracking algorithms to ROS.
Jan 2019 - Jul 2019
Aachen, Germany
Indurad
Worked in the IT department, developing VPN solutions and creating debugging software for automatic equipment maintenance across various sites.
Jun 2018 - Dec 2018
Aachen, Germany

Publications

Learning 3D-Gaussian Simulators from RGB Videos
M. Zhobro, A. Geist, G. Martius
Train simulators from multi-view videos by leveraging a 3DGS particle set for state representation and maintaining an inductive-bias-free dynamic model using the TEM-PTV3 transformer architecture.
Arxiv, 2025
Learning with 3D rotations, a hitchhiker’s guide to SO(3) Guidance on selecting the right rotation representations on deep learning models.
ICML, 2024