Moritz Oster 🐝

he/himMoritz Oster

Researcher

Deutsches Forschungszentrum
für Künstliche Intelligenz

Saarland University

Universität des Saarlandes

Campus D3 2, Room +2.10

66123 Saarbrücken

Germany

I am a researcher in the Neuro-Mechanistic Modeling group, working on sequential decision-making at the intersection of deep reinforcement learning and classical planning. I believe that neuro-symbolic approaches are the way forward: they combine the efficiency of learned models with the guarantees of symbolic methods, which purely neural approaches cannot offer.

In my Master’s thesis, I combined supervised pretraining on optimal planning data with reinforcement learning fine-tuning to learn per-domain generalizing policies. I joined the group as a full-time researcher in June 2026.

Projects

  • preAIrranging – Preparing an AI-Based Rearranging Hub (abat+, DFKI, Saarland University): we use neuro-symbolic approaches to solve the combinatorial optimization problems faced in modern production processes in industrial manufacturing.

Interests

I am interested in reinforcement learning and classical planning, especially in combining the two. Outside of research, I love cooking, doing sports, traveling, and I keep bees.
  • Reinforcement Learning
  • Classical Planning
  • Deep Learning
  • Combinatorial Optimization

Education

  • M.Sc. Computer Science, 2026

    Supervised by Prof. Dr. Verena Wolf at Saarland University

    Uniting Supervised and Reinforcement Learning for Per-Domain Generalizing Policies

  • B.Sc. Computer Science, 2024

    Saarland University

Publications (2)

2026

1 paper

Per-Domain Generalizing Policies: On Learning Efficient and Robust Q-Value Functions

Nicola J. Müller, Moritz Oster, Isabel Valera, Jörg Hoffmann and Timo P. Gros

2026 — Proceedings of the Thirty-Sixth International Conference on Automated Planning and Scheduling. International Conference on Automated Planning and Scheduling (ICAPS-2026), June 27 - July 2, Dublin, Ireland

@inproceedings{mullerPerDomainGeneralizingPoliciesOn2026,
 author = {Müller, Nicola J. and Oster, Moritz and Valera, Isabel and Hoffmann, Jörg and Gros, Timo P.},
 booktitle = {Proceedings of the Thirty-Sixth International Conference on Automated Planning and Scheduling. International Conference on Automated Planning and Scheduling (ICAPS-2026), June 27 - July 2, Dublin, Ireland},
 doi = {10.1609/icaps.v36i1.42889},
 isbn = {ISSN 2334-0843},
 month = {6},
 number = {1},
 pages = {703-707},
 publisher = {AAAI Press},
 title = {Per-Domain Generalizing Policies: On Learning Efficient and Robust Q-Value Functions},
 url = {https://ojs.aaai.org/index.php/ICAPS/article/view/42889},
 volume = {36},
 year = {2026}
}
DOI URL

2025

1 paper

Learning Per-Domain Generalizing Policies Using Offline Reinforcement Learning

Nicola J. Müller, Moritz Oster and Timo P. Gros

2025 — Proceedings of the Ninth Workshop on Bridging the Gap Between AI Planning and (Reinforcement) Learning (PRL @ ICAPS 2025). Workshop on Bridging the Gap Between AI Planning and Reinforcement Learning (PRL-2025), located at ICAPS'25, November 10, Melbourne, Australia

PDF 
@inproceedings{mullerLearningPerDomainGeneralizingPolicies2025,
 author = {Nicola J. Müller and Moritz Oster and Timo P. Gros},
 booktitle = {Proceedings of the Ninth Workshop on Bridging the Gap Between AI Planning and (Reinforcement) Learning (PRL @ ICAPS 2025). Workshop on Bridging the Gap Between AI Planning and Reinforcement Learning (PRL-2025), located at ICAPS'25, November 10, Melbourne, Australia},
 month = {11},
 publisher = {AAAI Press},
 title = {Learning Per-Domain Generalizing Policies Using Offline Reinforcement Learning},
 url = {https://prl-theworkshop.github.io/prl2025-icaps/papers/11.pdf},
 year = {2025}
}