Design And Implementation Of Pallet Identification And Positioning System Based On AGV

Automated Guided Vehicle(AGV)overcomes the problems of low processing efficiency and high labor cost in traditional logistics technology hence is widely used in the field of intelligent logistics.How to make the process of identification and positioning pallets by AGV more efficient and real-time has become the research hotspot.To address this challenge,in this thesis,we analyzed the operation requirements of AGV in detail,employed the depth camera to obtain the 3D point cloud of the target pallet,and designed new pallet identification and positioning methods.In addition,the AGV software and hardware platform are also built to verify the effectiveness of the proposed methods,which further improves the accuracy of pallet identification and positioning in complex scenes.Firstly,for the problem of pallet identification and detection in complex environments,we proposed a grid-based ground point filtering method to filter out the ground point noise after point cloud preprocessing.Subsequently,HSV color space,KD-Tree optimized Euclidean clustering and random sampling consensus design segmentation algorithm are combined to achieve target region segmentation,and then pallet identification is completed based on morphological operations and pixel point calculations.Secondly,we proposed a point cloud hole repair method based on the meshing idea to address the problem of recognition failure caused by holes in the point cloud plane,and the mesh intersections are used as new added data points to achieve hole repair.Meanwhile,the pose information of the target pallet is obtained by calculating the feature points and feature line segments.Depth map feature matching is used to obtain feature points for motion model estimation.The obtained relative poses are used to fill in the missing information after recognition failures to achieve pallet localization.Finally,the AGV experimental platform is designed and built.The sufficient experimental results show that the proposed pallet identification and positioning algorithm has high real-time performance and stability,and can be applied to most pallets in the market,and applied in engineering projects,which improves the intelligent process of AGV logistics handling and reduces the difficulty and cost of project implementation.