Researcher

Udo Frese

Udo Frese is a robotics researcher whose work spans computer vision, sensor fusion and SLAM for autonomous robots. He studied computer science at Paderborn, worked on SLAM at DLR, earned his doctorate at Erlangen-Nuremberg, and published Treemap for robot mapping. At the University of Bremen, he became a professor and his group runs RoboCup team B-Human; his later work covers assistive robot control, machine-learning near-fall detection and AI for lunar robotics. As of October 2026, he is a professor for multi-sensor interactive systems at Bremen and is associated with DFKI.

Timeline

  1. 1993–2004

    Paderborn, DLR and a doctorate

    He studies computer science at the University of Paderborn from 1993 to 1997. From 1998 to 2003, he works on simultaneous localization and mapping (SLAM) at the German Aerospace Center (DLR), and he defends his doctorate at Friedrich-Alexander-Universität Erlangen-Nürnberg in 2004.

  2. 2006–2007

    Treemap for robot mapping

    In 2006, he coauthors a paper with Lutz Schroeder on closing a million-landmark SLAM loop. In 2007, he publishes Treemap as a generic backend for six-degree-of-freedom SLAM, an approach that performs incremental Gaussian inference for robot mapping.

  3. 2008–2014

    Assistant professor at Bremen

    In 2008, he becomes assistant professor for real-time computer vision at the University of Bremen. In 2014, he becomes professor for multi-sensor interactive systems.

  4. 2024

    Adaptive control for assistive robots

    In 2024, he coauthors a study of adaptive shared control for assistive robots with people who have upper-limb impairments. It analyzes 24 participants across 81 trials using three input devices.

  5. By 2025

    B-Human in RoboCup

    By 2025, his group and DFKI run B-Human, a robot-soccer team that has won the RoboCup World Championship 11 times.

  6. 2025

    Machine learning for near-fall detection

    In 2025, he coauthors a Sensors paper on real-world near-fall detection with advanced machine learning.

  7. September 2026

    AI for planetary exploration

    In 2026, he coauthors a paper titled Artificial Intelligence for Planetary Exploration: Lessons Learned from a Decade of Analog Field Tests. In September 2026, he coauthors a paper on hardware-accelerated instance segmentation for space robotics; its team reports 309-millisecond inference on a lunar micro-rover at 5.7 watts.

  8. As of October 2026

    Professor at Bremen

    He is a professor for multi-sensor interactive systems at the University of Bremen and is associated with DFKI. His research interests include computer vision, probabilistic sensor fusion and safety algorithms for robotics.

Career

  1. Since 2008University of BremenAssistant Professor for Real-Time Computer Vision
  2. 1998–2003German Aerospace Center

Education

  1. 2004Friedrich-Alexander-Universität Erlangen-NürnbergDoctorate
  2. 1997University of PaderbornComputer Science

Sources and credits

Photo: We have no published photo of Udo Frese yet. Send us a free-licensed photo.

  1. en/UdoFrese - AG Multisensorische Interaktive Systeme informatik.uni-bremen.de. Checked 20 October 2026.
  2. TreeMap openslam-org.github.io. Checked 20 October 2026.
  3. UdoFresePublications - AG Multisensorische Interaktive Systeme informatik.uni-bremen.de. Checked 20 October 2026.
  4. Adaptive Control in Assistive Application -- A Study Evaluating Shared Control by Users with Limited Upper Limb Mobility arxiv.org. Checked 20 October 2026.
  5. Udo Frese | Faculty of Science and Engineering | University of Groningen rug.nl. Checked 20 October 2026.
  6. Hardware-Accelerated Instance Segmentation for Resource-Constrained Space Robotics with Criticality Analysis arxiv.org. Checked 20 October 2026.