Leonard Franz

Ph.D. Student · University of Tübingen

leonard.franz@student.uni-tuebingen.de GitHub

About

I am a Ph.D. student in the Autonomous Learning group at the University of Tübingen, advised by Georg Martius. My research focuses on reinforcement learning for musculoskeletal control — combining constrained RL, generative modelling, and biomechanical simulation to build physically realistic models of movement.

Research Interests

Publications

Drifting Fields are not Conservative
Leonard Franz†, Sebastian Hoffmann†, Georg Martius
arXiv 2026 arXiv preprint
† Equal contribution
Stochastic Decision Horizons for Constrained Reinforcement Learning
Nikola Milosevic†, Leonard Franz†, et al., Pavel Kolev
arXiv 2026 arXiv preprint
† Equal contribution

Experience

2023 – 2024
Research Intern
MPI-IS · Autonomous Learning, Tübingen
Constrained reinforcement learning for energy-efficient bipedal locomotion with musculoskeletal models.
2023 – 2024
Student Assistant
MPI-IS · Perceiving Systems · in cooperation with University Hospital Tübingen
Development of a virtual reality environment in Unity for the study of body image disturbance via eye tracking.
2024
Student Exchange
Monash University, Melbourne, Australia

Education

2025 – present
Ph.D. Machine Learning
Eberhard-Karls-Universität Tübingen · Autonomous Learning Group
2022 – 2025
M.Sc. Machine Learning GPA 1.2
Eberhard-Karls-Universität Tübingen
Thesis: Constrained Reinforcement Learning for Energy Efficient and Natural Locomotion in Musculoskeletal Systems
2018 – 2022
B.Sc. Physics GPA 1.4
Albert-Ludwigs-Universität Freiburg
Thesis: Feature Selection for Allosteric Processes

Projects

PloTwist
Python library for easy creation of HTML reports containing Matplotlib plots.
DawTTY
Python-based digital audio workstation for the terminal.

Skills

Languages German (native), English (C1), Portuguese (fluent), Spanish (basic) ML / Science PyTorch, JAX, Keras, NumPy, Hugging Face, SciPy, scikit-learn, Tonic RL Programming Python, C++, C#, Julia