Autonomous Cruise Unmanned Ship System

China has a large number of artificial lakes and hydropower lakes.With the rapid development of society and industrialization,the environmental pollution on the lake is becoming increasingly serious.In the era of focusing on ecological environmental protection,the issue of treating lake pollution has begun to be on the agenda.The traditional method of manually fishing for floating garbage on the water surface not only has a large workload,but also has certain safety problems.In order to solve this problem,this paper proposes an unmanned ship cruising along a designated route on the water surface,during the cruise,the ship can identify floating garbage on the water surface and track and salvage.It can not only play a role of security patrol on the lake,but also solve the problem of traditional methods of salvaging garbage,and realize the intelligence and efficiency of unmanned ships.This article studies the autonomous cruise of the unmanned ship and the salvage of garbage during the cruise,and conducts design tests on the unmanned ship platform.The specific work is as follows:(1)For the inflexible control of a single propeller-powered unmanned ship,a two-propellerdriven unmanned ship is used as the experimental platform in this paper.The motion control unit of the ship is analyzed based on the differential motion model.(2)The core technology of autonomous cruise is path planning.This paper implements path planning based on water map,GPS and IMU fusion pose and lidar.Due to the wide water area,the 2D scan data scanned along the shore using 3D lidar is used to create maps using cartographer-SLAM mapping algorithm based on graph optimization,and then adjusted with the help of GIMP software.In order to achieve autonomous cruise experiments for unmanned ships,it is necessary to combine global and local path planning for dynamic path planning of the hull.First,a path planning experiment for a single target point is completed,and then an autonomous cruise experiment design for a designated route is completed.(3)In order to salvage common rubbish during cruise,the key is target recognition and tracking.The accuracy and scalability of traditional target recognition methods are poor.With the development of deep learning technology,its application in the surface environment for garbage target recognition becomes possible.After obtaining the monocular image data on the unmanned ship in this paper,the deep learning method based on the yolov3-tiny model is used to identify and locate the floating garbage on the water surface,and finally the target is tracked and salvaged.