As part of the cooperative project Grainlab with TU Ilmenau (2014/2015), a compact laboratory automation system was developed for sorting wheat grains. The goal was to reliably filter out so-called contaminants – unwanted components such as foreign grains or damaged grains – from the wheat.
The process began with filling the wheat into a funnel. From there, the grains were evenly distributed onto five conveyor belts via a shaking plate. A line camera captured the grains during movement, and an early form of AI-supported image recognition, developed at TU Ilmenau, analyzed the images in real-time.
- Good grains were collected in a central container.
- Contaminants were ejected into separate containers depending on their type.
- Grains that were not clearly identified in the first pass were reintroduced into the funnel via a recirculation system.
I was responsible for the PLC programming with TIA Portal 7.4. The PLC controlled, among other things:
- the converters for controlling the conveyor belts,
- the ejection mechanism,
- as well as the communication with the camera and the user interface.
The user interface (UI) developed by me in LabVIEW enabled comfortable operation:
- The conveyor belt speed could be adjusted flexibly.
- The most recently captured grains were visualized in tracks.
- A software switch allowed the ejection to be disabled in order to specifically collect training data for the AI.
The PLC continuously transmitted the current belt increment to the user interface, so the associated Windows program always knew the position of the grains. At the same time, the current belt position was passed on to the camera to determine the optimal time for image capture.