Ozgur S. Oguz

Assistant Professor of Computer Engineering, Bilkent University · Director, Learning for Intelligent Robotic Agents (LiRA) Lab

bio_oz.webp

Office: EA-529, Engineering Bldg.

Dept. of Computer Engineering

Bilkent University, 06800 Ankara

I am an Assistant Professor in the Department of Computer Engineering at Bilkent University, where I lead the Learning for Intelligent Robotic Agents (LiRA) Lab.

Previously, I was a postdoctoral researcher at the Max Planck Institute for Intelligent Systems and the Uni. of Stuttgart (Learning and Intelligent Systems / Machine Learning & Robotics Lab) working with Marc Toussaint, affiliated with the Cluster of Excellence: IntCDC. I received my Ph.D. (summa cum laude) from the Technical University of Munich (TUM). I studied Computer Science at the Uni. of British Columbia (UBC) and Koç University.

Research Interests & LiRA Lab

My research sits at the intersection of Robotics, Machine Learning, and Artificial Intelligence. At the LiRA Lab, we specialize in building active, embodied AI agents capable of perceiving, reasoning about, and interacting seamlessly with the physical world:

  • Physical Intelligence & Embodied AI: Grounding vision-language models (VLMs) for dexterous manipulation.
  • Multi-Arm, Dexterous & Contact-Rich Control: Optimization-augmented reactive control and dynamic tool manipulation.
  • Task and Motion Planning (TAMP): Logic-Geometric Programming (LGP), sequential manipulation, and hybrid planners.
  • Robot Learning: Deep reinforcement learning, imitation learning, and generative policy synthesis.

Prospective Students & Researchers

  • Post-Doc, Ph.D., and M.Sc. positions are available in the LiRA Lab.
  • Highly motivated undergraduate and graduate students interested in embodied AI, robot learning, deep perception, and physical intelligence are encouraged to reach out via email with their CV and transcript.

news

Oct 05, 2026 Two new preprints on diffusion policies are out: P3: Persistent Particle Planning for Constrained Diffusion Control and Reachability-Aware Diffusion Policy Optimization.
Sep 14, 2026 26-Fall: CS 449 / 549: Learning for Robotics course - course page
Jul 01, 2026 Our paper Interpretable Responsibility Sharing as a Heuristic for Task and Motion Planning is published in Robotics and Autonomous Systems.
Jun 01, 2026 Our paper Uncertainty-Aware Safety Propagation Critics for Safe Reinforcement Learning is accepted to TMLR.
May 31, 2026 Our work Lagrangian Perturbation Diffusion Steering: Latent Reinforcement Learning for Generative Policies is accepted to ICML 2026.
May 04, 2026 Our work CoRAL: Contact-Rich Adaptive LLM-based Control for Robotic Manipulation is accepted to RSS 2026.
Mar 12, 2026 New preprint: Push, Press, Slide: Mode-Aware Planar Contact Manipulation via Reduced-Order Models.

selected publications

  1. 26-demiray-TMLR
    USPC_representative.png
    Uncertainty-Aware Safety Propagation Critics for Safe Reinforcement Learning
    Kutay Demiray, Efe Eren Ceyani, and Ozgur S Oguz
    Transactions on Machine Learning Research, 2026
    Bib PDF
  2. 26-cicek-RSS
    CoRAL_representative.png
    CoRAL: Contact-Rich Adaptive LLM-based Control for Robotic Manipulation
    Berk Cicek, Mert K Er, and Ozgur S Oguz
    RSS 2026 Robotics: Science and Systems 2026, 2026
    arXiv Bib PDF
  3. 26-simsir-ICML
    lagrangian_representative.png
    Lagrangian Perturbation Diffusion Steering: Latent Reinforcement Learning for Generative Policies
    Hikmet Simsir, and Ozgur S Oguz
    ICML 2026 Forty-Third International Conference on Machine Learning, 2026
    arXiv Bib PDF
  4. 26-yenicesu-RAS
    26-yenicesu-RAS.png
    Interpretable Responsibility Sharing as a Heuristic for Task and Motion Planning
    Arda Sarp Yenicesu, Sepehr Nourmohammadi, Berk Cicek, and 1 more author
    Robotics and Autonomous Systems, Jul 2026
    arXiv Bib PDF
  5. 26-tuncer-ICRA
    Mo-Segman_representative.png
    MO-SeGMan: A Rearrangement Planning Framework for Multi-Objective Sequential and Guided Manipulation in Constrained Environments
    Cankut Bora Tuncer, Marc Toussaint, and Ozgur S Oguz
    In 2026 IEEE International Conference on Robotics and Automation (ICRA), Jul 2026
    arXiv Bib PDF
  6. 25-tuncer-IROS
    SeGMan_representative.png
    SeGMan: Sequential and Guided Manipulation Planner for Robust Planning in 2D Constrained Environments
    Cankut Bora Tuncer, Dilruba Sultan Haliloglu, and Ozgur S Oguz
    In 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Jul 2025
    arXiv Bib PDF
  7. 25-cicek-RAL
    H-Map_representative.png
    H-MaP: An Iterative and Hybrid Sequential Manipulation Planner
    Berk Cicek, Arda Sarp Yenicesu, Cankut Bora Tuncer, and 2 more authors
    IEEE Robotics and Automation Letters, Jul 2025
    arXiv Bib PDF
  8. 22-braun-IROS
    22-braun-IROS.png
    RHH-LGP: Receding Horizon And Heuristics-Based Logic-Geometric Programming For Task And Motion Planning
    Cornelius V. Braun, Joaquim Ortiz-Haro, Marc Toussaint, and 1 more author
    In IEEE IROS, Jul 2022
    arXiv Bib PDF
  9. 22-hartmann-TRO
    22-hartmann-TRO.png
    Long-Horizon Multi-Robot Rearrangement Planning for Construction Assembly
    Valentin Noah Hartmann, Andreas Orthey, Danny Driess, and 2 more authors
    IEEE TRO - Transactions on Robotics, Jul 2022
    arXiv Bib PDF
  10. 21-schubert-NeurIPS
    21-schubert-NeurIPS.png
    Learning to Execute: Efficient Learning of Universal Plan-Conditioned Policies in Robotics
    Ingmar Schubert,  Driess, Danny, Ozgur S. Oguz, and 1 more author
    NeurIPS - Conf. on Neural Information Processing Systems, Jul 2021
    arXiv Bib PDF Code