Research On Warehouse AGV Based On Compound Navigation

As an important production and transportation tool,intelligent transportation robot has been favored by manufacturing and logistics industry.However,its development has been restricted by some problems such as AGV's navigation accuracy,trajectory flexibility and the success rate of manipulator grasping in the field.In view of the above problems,a compound navigation storage AGV is proposed under the background of warehouse storage,which has the functions of autonomous navigation,target recognition,object positioning,grasping and so on.The main research contents are as follows:(1)The mechanical structure of AGV trolley and manipulator was designed.The two wheel differential structure was adopted by AGV,which has two load-bearing universal wheels at the front and rear respectively.The type of palletizing and array suction cups mounted at the end are adopted by the manipulator.At the same time,the kinematics modeling of AGV was analyzed and the forward and inverse kinematics of manipulator was calculated by D-H method.(2)The modes of conventional navigation and composite navigation based on vision and RFID are introduced.First of all,the grid map was established by embedding the RF card in the landmark picture,and the coordinate information in the RF card was read by the RFID sensor to obtain the position of AGV.At the same time,the visual sensor was used to capture the landmark position to compensate the motion trajectory,so as to realize the navigation mode in coordination of flexibility and accuracy.(3)Through the research on grasping objects by manipulators,a method of identifying and locating objects is proposed.Firstly,YOLOv4-tiny target detection algorithm was used to identify the objects which waiting to be captured.Secondly,objects are divided into larger object and smaller object according to the size of the object.After that,stereo matching based on the these two situations,where Larger objects were located by SGBM algorithm and Smaller objects were segmented by PSPnet network.Finally,the manipulator was used to complete the grasping action.According to the above main functions,the storage AGV with composite navigation was independently designed for experimental verification.The experiments shows that the trajectory error,angle error and capture success rate of AGV navigation can meet the actual needs.