PhD defense of Sören Michalik on fast image processing for SLAM and mobile robotics

Congratulations to Sören Michalik, who has sucessfully defended his thesis entitled „Hardware-beschleunigte Bildmerkmale mit Subpixel-Genauigkeit zur SLAM Lokalisierung und Objekterkennung“ (in German, Englisch translation: Hardware-accelerated Image Features with Sub-pixel Accuracy for SLAM Lokalization and Object Recognition).
The thesis proposes a new set of sub-pixel accurate features detectors that are designed and implemented with an emphasis on hardware-acceleration through GPU and in particular FPGA adaptors. The features and their integration in SLAM are benchmarked on standard data-sets and show an significant increase in performance while achieving very high frame rates (~40 fps) at low energy consumption in the FPGA case, which makes them suitable for the application in all fields of mobile robotics or autonomous navigation. The thesis was evaluated by the examination commitee including Prof. T. Deserno at the helm and reviewers and supervisors Prof. M. Berekovic and Prof. J. Steil.