The “SpAss” project is developing an intelligent assistance system for recreational boating. It supports boat operators in challenging situations such as passing through locks, mooring and casting off, or whilst in port, thereby improving safety and navigation. By combining various sensors and utilising artificial intelligence, the system analyses the surroundings in real time, detects objects and provides specific recommendations for action. Data fusion technology and a central processing unit ensure high positional accuracy and enable precise trajectory planning – whilst utilising cost-effective hardware.

As part of the project, a head model is to be developed that is based both on CT data (3D volume data) from radiotherapy planning and on 2D optical images. The detection of the model parameters of the underlying head model is AI-based. Extracted keypoints (facial landmarks) of the head are used to generate a trackable pose. With the help of photogrammetric methods, a redundant algorithm can be developed that makes the AI methods used verifiable. In addition, conclusions can be drawn about the geometric registration between external and internal localised tumour positions. The aim is to generate reproducible patient positioning without additional stress in order to increase the precision of the therapy. This approach is intended to ensure the anatomical correctness and individual description of the patient so that treatment can be carried out in a (semi-)upright position with less strain on the patient and with fewer aids.