Towards Underwater Docking Through Teaching-by-Showing

We propose an underwater docking approach based on teaching-by-showing. A visual trajectory is obtained from a manual docking, then used as a reference during autonomous mode. Using a full trajectory instead of a single reference at the docking pose allows for more variation in the initial pose. It is also compatible with a moving docking station, as the vehicle will adapt its motion to track the trajectory in the image. A proof-of-concept is illustrated in a simulation scenario.