Abstract: Objectives Autonomous recovery of Unmanned Surface Vehicle (USV) using docking device under complex sea conditions faces the influence from natural wind and waves, mother ship wake and harsh weather conditions. How to robustly sense the guidance information of the docking device and realize USV motion target tracking is an urgent problem to be solved. Methods This paper proposes a docking guidance strategy based on multi-source observation of the docking device. The whole recovery process is divided into three stages. In the first stage, the USV avoids the mother boat's wake interference zone and completes the rough alignment of the guidance route. In the second stage, the USV maintains the heading tracking guide line in preparation for the end docking recovery. In the third stage, the data obtained from the camera sensor and the inertial navigation sensor are filtered and fused and passed to USV to calculate the guidance line. And then USV completes the task of tracking and return to the docking device. Results In this paper, an USV docking recovery system based on visual and inertial sensor fusion technology is proposed. The feasibility of system