Research On Bin-Picking Algorithms And System Based On 3D Sensor For Free-Form Objects

With the aging population and the improvement of intelligent technology,the industrial robot market and industry have expanded rapidly.As a typical mechanical device in the field of robots,the manipulator has a wide range of applications in industrial,medical,entertainment and other fields.Grabbing is the most important task performed by the manipulator,and it is a key technology in robotic automation processes such as sorting,assembly,loading,welding,spraying,and surgery.In industrial,medical and other fields,there are high requirements for the robustness,grab accuracy and adaptability to complex environments.This article takes scattered and stacked small objects as research objects,focusing on the construction of the manipulator's grasping system,3D object reconstruction,point cloud recognition,and positioning,and control technology of manipulator,a 3D vision-guided bin-picking system for free-form targets was proposed to complete reliable,fast,and accurate grasping.The main research points and major findings of this article can be concluded briefly below: Aiming at the industrial grasping requirements of scattered small objects,a stable and reliable manipulator bin-picking system was established,the vision system based on the structured light of the fixed platform and the vision system based on the line structured light of the transfer platform is designed.The system can be divided into stereo vision module,point cloud processing module and grab control module.In the stereo vision module,the method of coded structured light and line structured light is used to realize 3D reconstruction,and a 3D template library is established.The point cloud processing module denoises point cloud and then uses a template matching algorithm to match the standard template with the scene point cloud.After obtaining the matching parameters,the hand-eye calibration parameters are used to calculate the gripping position and attitude of the manipulator.According to the reliable,fast and accurate grasping work of the manipulator,this paper constructs a structured light system,utilizing the advantages of high accuracy and fast reconstruction of the structured light system,high-precision 3D information of the object is obtained.In order to accurately calculate the pose of the target,the point cloud is filtered,and then 3D point matching algorithm of point-to-feature is used to accurately match the scene and the model point cloud.The target pose can be calculated by the coordinate system transformation so that the robot can correctly grasp the target.The system can grasp targets in an accurate position in scattered objects,and reliably,quickly and accurately complete the grasping work of manipulator.It can be used in industrial automation fields such as sorting,assembly,and loading,especially in the field of small target grasping,to improve the performance of existing production lines.