Research On Charging Interface Recognition And Pose Acquisition Algorithm Based On 3D Point Cloud

With the popularity of new energy electric vehicles and the increasing market scale,the problem of automatic charging of electric vehicles has gradually become a research hotspot.At the present stage,the plugging and unplugging of electric vehicles is mainly done by human resources.In some commercial and public places,logistics and warehousing,where there are more electric working vehicles or electric passenger cars,an "automatic charging robot" is urgently needed to complete the complicated plugging and unplugging work to improve charging efficiency and save equipment and management costs.For the automatic charging robot,the identification and positioning of the vehicle charging port is an extremely critical part of the stable operation of the charging robot.Therefore,this project analyzes the problem of low recognition accuracy,inaccurate positioning and slow speed of the current charging port,and completes the following research:(1)Software and hardware platform construction.The problems in the existing charging port identification process were analyzed,and the three-dimensional point cloud of the charging port was established as the identification object.And the charging robot experimental platform was built,which contains high-precision 3D camera,UR5 robot,computer and other equipment,and then the software development environment was introduced.And the point cloud image acquisition was carried out with the AC charging port module specified by the current latest national standard GB/T 20234.2-2015 as the research object.(2)Charging port identification study.The acquired point cloud is pre-processed,mainly including point cloud downsampling,point cloud denoising,feature analysis of the point cloud image,using density-based clustering segmentation,conditional filtering and other operations to separate the point cloud of the charging port.(3)Study of charging port pose acquisition.The separated point cloud is further analyzed and processed,and the existing matching algorithm is used to explore and compare the verification,and problems such as large matching angle error and poor robustness are found.The geometric feature-based charging port pose acquisition algorithm is proposed to gradually disassemble and identify the external end face and internal jack of the charging port and extract its position information;according to the distribution characteristics of the charging port point cloud using plane fitting calculation,normal vector calculation,double enclosing frame extraction method,and combined with space vector operation to extract its direction information.The confidence degree is calculated by using both internal and external pose information,and the position and pose information are integrated according to the pose information determined by the confidence degree to finally obtain the charging port pose matrix.(4)Charging port attitude verification experiment.The hand-eye calibration of the eyein-hand model is completed by combining the spherical center calibration method,and the chisquare transformation matrix of the camera coordinate system and the robot arm base coordinate system is obtained.And the charging port recognition upper computer software was developed in combination with the subject task,which can realize point cloud acquisition,point cloud preview,point cloud interaction,posture visualization,posture data output,robotic arm control and other functions that are convenient for debugging.Finally,the experiment of multiple sets of positional recognition records,the control experiment of real positional and recognized positional,and the experiment of charging port docking have been completed,and the positional verification work has been completed.The position error of the recognized posture is 0.61 mm and the angle error is 0.63°,which can realize the robot charging docking.The experiment shows that this study has high real-time,reliability and accuracy,and has practical application value.