Abstract: Objectives Aiming at the real-time acquisition of the relative position and attitude between the mother ship and the unmanned surface vehicle (USV) during autonomous recovery operations, this paper proposes a LiDAR-based guidance method for autonomous USV recovery. Methods First, perform real-time 3D object detection between the mothership and USV using the PointPillars algorithm to obtain their relative pose. Then, establish a target tracking framework with Kalman filtering and Hungarian algorithm to filter pose noise and achieve temporal association of detection results, ensuring stable motion state output. Finally, calculate heading deviation via line-of-sight guidance algorithm to drive USV recovery. Results In the adaptability evaluation experiment of the 3D object detection algorithm, the position detection error of the USV was 0.0712 meters, and the heading detection error was 1.518 degrees. In the autonomous recovery simulation experiment of the USV, all centering errors were less than 0.6 meters. Conclusions The feasibility of a single LiDAR-based recovery guidance method has been validated, offering a novel solution approach for autonomous recovery guidance of USV.