The boundary between mechanics, electronics and software —automation, robotics, computer vision and intelligent control or industry 4.0
— has fascinated me from an early age. I ended up studying mechanical engineering because experimenting with software development and electronics at home was quite easy while my possibilities to experiment with mechanics were rather limited at that time.
As an engineer, I love innovative and challenging projects in this area in which I can combine my experience as a software engineer with my interest for and knowledge of mechanical systems, electronics, physics and mathematics and where I can combine proven solutions and scientific innovations. I endeavour a solution that makes you happy and that I can be proud of.
Since 2017, I'm a partner at Kapernikov. We specialize in data intensive and machine vision applications for asset management and industry and have a wonderful team of more than thirty smart and highly skilled engineers to help our clients tackle their challenges.
Master of Science in Engineering (KU Leuven, 2003)
Doctor of Engineering (KU Leuven, 2009)
And lots of tinkering with interesting technologies…
Languages: C, C++, Elixir, Erlang/OTP, Fortran, Python
Frameworks and libraries: Qt, ROS (1 & 2)
Mathematics: Matlab, NumPy, Octave, SciPy
Machine learning: Dlib, Keras, Scikit-learn, TensorFlow
Computer vision: MVTec Halcon, OpenCV, PCL (Point Cloud Library)
Databases: Berkeley DB, MySQL, PostgreSQL, SQLite
Design: AutoCAD, Inventor, Creo (Pro/ENGINEER)
Simulations: Nastran & Patran, Creo (Pro/ENGINEER)
I developed a machine learning algoritm for offline configuration and online adjustment of optical sorting machines. I used LIBLINEAR for maximum performance, both for training and prediction and integrated the algorithm in the client's embedded C++ application.
I codeveloped the computer vision part of a quality control system performing 2D and 3D (using laser triangulation) tests on finished products. We used the ROS framework with Python and C++ and OpenCV for common computer vision algorithms.
I codeveloped a computer vision system to improve the effectiveness and efficiency of mobile material handling robots. We used the ROS framework with C++ and Python, OpenCV for common computer vision algorithms, Dlib and TensorFlow for 2D object detection and classification, CasADi and NOMAD for mathematical optimisation and a time-of-flight camera and PCL to verify the 3D reconstruction.
I joined the Kapernikov team in the 2017 and 2018 ArcelorMittal #hack4steel hackathons.
During the 2017 edition, we designed and implemented an AI system that, based on the material and image data provided by ArcelorMittal, predicts whether a weld would break or not and a small game to train operators in interpreting the currently available image data. Our solution ranked second-best of all teams in the competition category Artificial Intelligence.
During the 2018 edition, we designed and implemented an application that identifies people in a camera stream, locates and tracks them on a map and checks whether they are using the required personal protective equipment, with which we won the award in the Vision AI category. Our solution was also presented and demonstrated on the november 2018 Microsoft Tech Summit in Brussels.
I codeveloped a demonstrator to automatically read the License Plates of Global Transport Labels (GTL) on incoming goods on passing forklifts using a computer vision system based on Matrox Imaging Library and Matrox Design Assistant.
I developed a model predictive ventilation controller, a simulation framework and a library to communicate with a KNX home and building control system in Erlang/OTP.
For my PhD research, I developed a generic non-deterministic (interval and fuzzy) finite element solver for transient and steady-state structural dynamic analyses. I used C++ for the core, calling MSC Nastran for deterministic finite element analyses, various Fortran libraries for matrix operations and Matlab or GNU Octave for mathematical optimisation and visualisation.