Manuel’s research in cognitive modeling and computer vision at the University of Tübingen culminated in the development of FLIP, a fovea-inspired vision transformer, earning him the ENNS Student Award at ICANN 2024.
Working in the field of cognitive modeling and computer vision at the University of Tübingen, Manuel's research on object-centric vision systems has earned him the ENNS Student Award at ICANN 2024 and produced a segmentation model that outperforms META's Segment Anything Model (SAM) with three orders of magnitude fewer parameters.
Manuel joined IMPRS-IS in April 2021, working in the Cognitive Modeling group at the University of Tübingen under the supervision of Prof. Martin V. Butz. His research focuses on developing advanced object reasoning systems inspired by the human brain, particularly through the LOCation and Identity tracking system (Loci) and its variants.
During his master's thesis, Manuel designed and 3D-printed a many-joint robotic arm controlled using Spiking Neural Networks, demonstrating successful control of arms with up to 75 degrees of freedom. This work, published at IROS, showcased fully modular joints at a hardware cost under €500 per robot.
Manuel's doctoral research has produced several contributions. His Loci system became the CATER tracking champion by disentangling "what" and "where" into separate codes. Building on this, Loci-Looped became the first slot-based system to learn object permanence, inertia, and solidity directly from raw videos without ground-truth information. His Loci-Segmented model extended this framework to more challenging datasets, earning him the ENNS Student Award at ICANN 2024.
Most recently, Manuel developed FLIP, a fovea-inspired vision transformer that outperforms META's Segment Anything Model while using three orders of magnitude fewer parameters. You can try this model at: https://cognitivemodeling.github.io/FLIP/#demo
Manuel holds a Bachelor's in Computer Science from Karlsruhe Institute of Technology and a Master's from the University of Tübingen. Beyond research, he is passionate about 3D-printing and robotics. In his free time, he is an avid strength athlete and outdoor enthusiast.
Currently wrapping up his doctoral thesis, Manuel will continue as a postdoctoral researcher in Prof. Martin V. Butz's Cognitive Modeling lab, further developing his vision models and their applications.
Manuel's evolution of 3D-printed robotic arms: from his initial master's thesis design to two subsequent iterations, that are featured in an IROS paper: https://ieeexplore.ieee.org/abstract/document/9636001
More information