Research on visual multi-target based pose estimation algorithm for ARV underwater docking

Underwater docking technology has important scientific significance and practical value. By using ARV to connect with the underwater docking station, on the one hand, the batteries in ARV can be recharged, and the continuous operation capability can be enhanced. On the other hand, the ARV can be used to transmit the data collected by the docking station for a long time. In this paper, a multi-target combined pose estimation algorithm which combined the guide lights and AR markers is proposed to estimate the pose of the ARV. A weight changes with the distance between the camera and the docking port is employed to combine the poses calculated from the AR markers and the lights. Through the pool experiment, the effectiveness of the algorithm is proved in aspects such as pose estimation accuracy, effective distance and estimation success rate.