The 47th Special seminar of the School of Engineering

The ability to self-localize and navigate in unfamiliar environments is essential for mobile autonomous robots. However, camera-based approaches, particularly those designed for low-cost service robots, often struggle in unstructured outdoor environments with natural vegetation.
This talk introduces the biologically inspired Slow Feature Analysis (SFA) and explores the application of holistic SFA for self-localization and mapping, demonstrating its robustness and long-term stability in dynamic environments.
The second part showcases the robot’s navigation capabilities within the SFA space, relying solely on learned representations.
Finally, the talk will discuss Hybrid SFA, which builds on semantic simultaneous localization and extends it to learning visual landmarks with minimal supervision and interaction.

Ute Bauer-Wersing earned a Computer Science degree with distinction from the University of Nuernberg-Erlangen and a PhD on visual cortex processing models at Bielefeld University and University College London. After postdoctoral research at TU Berlin, she became a professor at Trier University of Applied Sciences in 2004 and joined Frankfurt University of Applied Sciences in 2007. Since 2012, she has led the Machine Learning, Computer Vision, and Autonomous Systems group and co-founded the PhD Center for Applied Computer Science. She has published over 30 papers, served on AI and robotics committees, and co-holds four patents. She also held leadership roles in IT consulting at Daimler-Chrysler and bioinformatics research at Proteome Sciences Plc, applying machine learning to laboratory automation and biomarker discovery.

• Ookayama West Area 20 West Bldg. 8W (Room W8-E1001)

• School of Engineering | Science Tokyo organization | About Science