Tim Walter

Doctoral Candidate, Technical University of Munich

About Me

I am a doctoral candidate in the Cyber Physical Systems Group at the Technical University of Munich, supervised by Prof. Matthias Althoff .

Before starting my PhD, I completed a M.Sc. degree in Computational Science and Engineering at TUM with my Master’s Thesis focusing on Leveraging Analytic Gradients for Provably Safe Reinforcement Learning . Prior I completed a dual study program with Rohde & Schwarz, where I completed an apprenticeship as an Electronics Technician for Information and Systems Technology, worked within the company for a cumulative time of 2 years, and earned a B.Eng. degree in Electrical Engineering and Information Technology from the University of Applied Sciences Munich, with my thesis focusing on outlier detection in manufacturing data for service predictions.

My current research interests lie in Reinforcement Learning, Modular Robotics and Algorithmic Design.

Publications

RAM: Reachability Across Morphologies

Preprint

RAM provides a fast, differentiable surrogate for pose reachability across morphologies, enabling rapid design and planning workflows.

Tim Walter, Xinyu Chen, Jonathan Külz, Matthias Althoff

Project Paper Video Code Dataset

@misc{walter2026ramreachabilitymorphologies,
      title={RAM: Reachability Across Morphologies}, 
      author={Tim Walter and Xinyu Chen and Jonathan Külz and Matthias Althoff},
      year={2026},
      eprint={2606.09108},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2606.09108}, 
}

Leveraging Analytic Gradients in Provably Safe Reinforcement Learning

IEEE Open Journal of Control Systems 2025

With SafeGBPO we developed the first effective safeguard for analytic gradient-based reinforcement learning. We analysed existing, differentiable safeguards, adapted them through modified mappings and gradient formulations, and integrated them into a state-of-the-art learning algorithm and a differentiable simulation.

Tim Walter, Hannah Markgraf, Jonathan Külz, Matthias Althoff

Project Paper Code

@article{11154003,
  author={Walter, Tim and Markgraf, Hannah and Külz, Jonathan and Althoff, Matthias},
  journal={IEEE Open Journal of Control Systems}, 
  title={Leveraging Analytic Gradients in Provably Safe Reinforcement Learning}, 
  year={2025},
  volume={4},
  number={},
  pages={463-481},
  doi={10.1109/OJCSYS.2025.3607845}}

NeRSemble: Multi-view Radiance Field Reconstruction of Human Heads

Siggraph 2023

NeRSemble reconstructs high-fidelity dynamic radiance fields of human heads. We combine a deformation for coarse movements with an ensemble of 3D multi-resolution hash encodings. These act as a form of expression-dependent volumetric textures that model fine-grained, expression-dependent details. Additionally, we propose a new 16 camera multi-view capture dataset (7.1 MP resolution and 73 frames per second) containing 4700 sequences of more than 220 human subjects.

Tobias Kirschstein, Shenhan Qian, Simon Giebenhain, Tim Walter, Matthias Nießner

Project Paper Video Code Dataset

@article{kirschstein2023nersemble,
    author = {Kirschstein, Tobias and Qian, Shenhan and Giebenhain, Simon and Walter, Tim and Nie\ss{}ner, Matthias},
    title = {NeRSemble: Multi-View Radiance Field Reconstruction of Human Heads},
    year = {2023},
    issue_date = {August 2023},
    publisher = {Association for Computing Machinery},
    address = {New York, NY, USA},
    volume = {42},
    number = {4},
    issn = {0730-0301},
    url = {https://doi.org/10.1145/3592455},
    doi = {10.1145/3592455},
    journal = {ACM Trans. Graph.},
    month = {jul},
    articleno = {161},
    numpages = {14},
}

Teaching

Practical

Provably Safe Reinforcement Learning Challenge

Instructor - Summer Semester 2026

This practical course focused on reinforcement learning with formal safety guarantees. Participants were provided with a set of RL environments including established baseline algorithms. Their task was then to innovate upon these baselines to improve the performance of the RL agent, while still maintaining the safety properties.

Practical

Safe Reinforcement Learning for Modular Robots

Instructor - Summer Semester 2026

Offered and supervised a project for a 4 student team on the following topic:

Morphology Optimisation using RAM and Newton (DORN)

Practical

Safe Reinforcement Learning for Modular Robots

Instructor - Winter Semester 2025/26

Offered and supervised projects for teams of 2-4 students on the following topics:

Differentiable Solver Unrolling for Safe RL
Differentiable Implicit Projections for Safe RL