Dynamic Scene Construction Based On Laser Three-dimensional Vision And Route Planning For Large Bulk Carrier

As a main equipment for material handling at the port,the bridge grab ship unloader has the advantages of adapting to the harsh operating environment of the wharf and low operating cost.In order to meet the demand for the development of efficient transportation of bulk materials,it is necessary to improve the manual operation mode of ship unloader to improve the automation of operation and reduce labor intensity.With advantages of high measurement accuracy,large measurement scale and strong anti-interference ability,laser measurement technology has been widely used in industrial measurement and autonomous navigation filed in recent years.In this paper,laser measurement is applied to the operation scene of ship unloader,and the degree of automation of ship unloader is improved based on laser three-dimensional vision technology.Automating ship unloader operations requires addressing two core issues: building and understanding of operation scenes and the path planning of grab,the main contents of this paper are as follows:In this paper,laser three-dimensional scanning technology is used to scan the working scene of ship unloader,and the 3D reconstruction of the scene is realized by using the scanned point cloud data.In the three-dimensional point cloud scene,the method of SHOT feature matching combined with 3D Hough transform is adopted to realize the recognition and localization of the reference reflector,and the position of the laser scanner in the world coordinate system of the working scene is further completed according to the relative position relationship.A method based on elevation statistics is proposed to segment the deck plane.A RANSAC algorithm based on the point cloud normal sphere feature is used to segment the cabin plane.Based on the segmented deck and cabin plane,the feasible working area of the grab is further determined.And point cloud of the material area can be segmented,further,material elevation distribution statistics information can be obtained by meshing the material area.Route planning of grab mainly studies the planning of grab grasping point,and the collision avoidance,anti-shake,and control strategy of discharging and unloading during the operation of grab.Based on the result of gridding of material area,the automatic planning strategy of grab grasping point is considered synthetically from the transverse and longitudinal directions,which respectively are the moving direction of cart and trolley;the collision avoidance strategy is proposed based on the feasible working area of grab.In order to analyze the anti-swing and material discharging strategy during the operation of the grab,the physical model of the grab-trolley system of the ship unloader was first established and the model was analyzed.A neural network method based on genetic algorithm optimization was proposed to realize the anti-swing control of the grab under the condition of variable rope length in the process of lifting and linkage of the grab-trolley system.In order to improve the efficiency of automatic operation of ship unloader,a way of discharging and unloading material is proposed,and the way of dynamic unloading is realized,and the anti-swing problem in the process of returning to the sea side after unloading is comprehensively considered.Based on Adams,the kinematic simulation of grab-trolley system was carried out to verify the effectiveness of the proposed path planning algorithm.Programmed and realized the unloader automatic operation control software based on VS and Qt and open source database PCL which is specially used for point cloud processing as a development tool,the automatic operation control software mainly includes the core functions of automatic operation of bulk carrier based on laser three-dimensional vision and dynamic planning of grabbing path,and providing a friendly man-machine interface.Finally,the core algorithm of segmentation of point cloud scene is tested by using field measurement data,and the test results verify the effectiveness of the algorithm.