Enhancement and Evaluation of an AUV's Docking Station Detection with Deep-Learning Based Object-Detection

Autonomous docking for underwater vehicles, especially locating the docking station, presents significant challenges for deploying sub-sea resident AUVs in exploration and monitoring tasks. To extend a fiducial marker-based docking station detection,  use of the state-of-the-art object detection deep learning models, specifically YOLOv8 in various sizes is proposed. The robustness of these models in detecting docking stations is assessed by training different model sizes under various configurations on a dataset collected at the DFKI test basin in Bremen, Germany. To show and improve their performances in a real-world under-ice scenario,  a previously recorded dataset from Torneträsk lake in Abisko, Sweden is utilized. The performance of these models is then compared to the previously used fiducial marker-based docking station detection. The result shows that a combination of both the classical detection method with one of the trained YOLOv8 models improves the detection performance significantly.