To provide a scalable alternative, this thesis develops and evaluates a computer vision-based software artifact designed to automatically detect, track, and spatially localize ground handling vehicles using a single-camera setup.

Following a design science research methodology, the system was trained on a custom video dataset collected at Zurich Airport, utilizing a YOLOv8 object detector, a procedural tracker, and planar homography for coordinate mapping. Evaluation results demonstrate that the system successfully identifies key apron vehicles and achieves high spatial accuracy, projecting real-world coordinates with a margin of error of less than ≈ 0.5 meters. However, while the artifact successfully generates spatial density patterns, the tracking component exhibited instability due to frequent occlusions, which currently limits the extraction of precise process timing data.

The research concludes that a computer vision system is a viable sensor for spatial monitoring but requires enhanced tracking logic to fully automate temporal process analysis.