Varroa mites are a major cause of bee mortality. Sebastian Steppuhn, a student from Pforzheim, created a device to identify and treat bees infested with these mites. His invention won him the main prize of 1500 euros at the Federal Artificial Intelligence Competition (BWKI) and an internship at the robotics company FANUC.
This year’s BWKI, organized by the Tübingen AI Center, had the theme “What freedom does AI give you?”. Ten teams presented their projects to the jury and the public. The event included a KI-Makerspace, Holobox, an interactive marble run, and an alumni meeting, making it a relaxed yet exciting occasion.
German President Frank-Walter Steinmeier attended and praised the young participants for their creativity and dedication, noting that many worked on their projects in their free time alongside school.
The BWKI, established in 2019 by researchers from the Tübingen AI Center, offers a free AI course for schools, with the possibility of being named “AI School of the Year”. The Tübingen AI Center is supported by the Federal Ministry of Education and Research and the Carl-Zeiss-Stiftung.
The Varroa mite, only 1.1 mm long and 1.6 mm wide, poses a significant threat to bees by attaching to their bodies and transmitting diseases. Without treatment, an infested bee colony can die within a few years. Typically, entire colonies are treated with formic acid, but this affects healthy bees as well.
Sebastian Steppuhn, a hobby beekeeper, developed an intelligent gateway for his beehive to identify and isolate infested bees for treatment. His system uses a camera and a machine-learning algorithm to detect mites. Initially, he faced challenges due to a lack of training data, so he created synthetic images of mites on bees to train the YOLO system. This allowed him to identify real instances in his videos and build a comprehensive dataset.
The system sorts bees using a gentle air mechanism to avoid harming them. In a hive of 25,000 bees, it identified 18 infested bees with 80 false positives, which is common in mass screenings. The benefit is treating fewer healthy bees and controlling mite spread.
Other award-winning projects included Leonie Weiss’s TrafficAid, an intelligent traffic light system that reduces CO₂ emissions by minimizing unnecessary stops. Anna Perkovic and Nicholas Dahlke explored why hot water freezes faster than cold in an ambitious experiment, using AI to classify frozen droplets. Lorenz Rutkevic’s project optimized microscopic cell images to aid medical diagnoses, and Peter Fuchs’s “FolderCopter” answered questions using a local language model.
Fabian Then, a 15-year-old, entertained with his smart fridge, which learned optimal storage for food items. It even corrected President Steinmeier, advising against refrigerating pineapple.