Positioning of an underwater vehicle based on model-known vision system

As a basic research project for Chinese 7000 m Manned Vehicle, positioning of an underwater vehicle through model-known vision system is developed and demonstrated. Model-known observed target is calibrated in advance and placed in the working space when positioning is demanded. Based on the vision system, accurate and reliable 3D pose information and 3 axis translations between the vehicle and observed target are estimated in real time. With Such information, positioning of the underwater vehicle is obtained. The pool experiment further demonstrates that the vehicle can move to a desired position and maintain a fixed position and orientation with the presence of man-made disturbance.