Research On Control Method Of Unmanned Surface Vehicle Back To Dock Based On Machine Vision

With the rapid development of the robot industry and ocean resource exploration and exploitation,the application of Unmanned Surface Vehicles(USVs)has become increasingly widespread.Autonomous docking technology is one of the key technologies for the autonomy and intelligence of USVs,but there are still many challenges and difficulties to be overcome.In this thesis,focusing on the docking task of small USVs,a dock with a guide pole structure was designed.Drawing from the characteristics of traditional planar targets,a more functional 3D target was designed.Based on machine vision technology,two different docking control methods were proposed:the visual heading trajectory-based docking control method and the trajectory tracking-based docking control method.The docking task was divided into three parts for research: the docking system,the dock detection system,and the docking control system.(1)In the dock docking system,the 3D target serves as a marker for the dock and guides the USV to the vicinity of the dock.The guide pole of the dock guides the nearby USV into the dock for precise docking.The feasibility of the system was verified through joint testing of the USV and the dock.(2)In the dock detection system,YOLOv5 object detection and Open CV image processing techniques were employed to detect the 3D target and extract target information.The feasibility of this system was validated through simulation experiments.In field experiments,using a 3D target with dimensions of 0.71 m × 0.36 m × 0.51 m,the detection distance of the system reached 11.7 m.(3)In the docking control system based on visual heading trajectory,the concept of visual heading trajectory was proposed based on the structure of the 3D target.The heading control task of the USV docking was converted into a visual heading trajectory tracking task,and the docking task was completed through the PID algorithm.The feasibility of the system was verified through simulation experiments,and the field experiment successfully completed the docking task with the USV at a distance of about 6 meters from the target.The system only requires one camera to complete the docking control,but is not suitable for environments with obstacles around the dock.(4)In the docking control system based on trajectory tracking,a positioning technology based on the 3D target was proposed,and pure pursuit algorithm was studied and applied to the trajectory tracking of the USV to complete the docking task.The feasibility of the system was verified through simulation experiments.The system requires one camera and an IMU sensor and can avoid obstacles by selecting a new docking reference trajectory.It is suitable for a wider range of environments than the first docking method.Through simulation experiments and some field experiments,the feasibility of the two docking control methods based on machine vision was verified.Both methods have the characteristics of low cost and high reliability,providing new solutions for USV docking technology.