Research On Workpiece Recognition And Pose Estimation For Industrial Robot Painting

Industrial robots have a widely use of applications in industrial automation,the emergence of spray robots has led to the rapid development of automatic spraying.Existing automatic spraying is mainly depend on path teaching and fixed tooling,which can get the spraying track and position of workpiece.But with the development of flexible manufacturing,traditional automatic spraying methods cannot adapt to production requirements for low-volume multi-type and no fixed fixtures,which mainly relies on manual work.However,the harsh spraying environment is harm to human health.Thus,it is of great significance to realize automatic spraying for workpiece without tooling.Against the above problems,this article researched a method for the recognition and pose estimation of spraying workpieces based on three-dimensional visual,using a line-structured sensor with strong adaptability to the environment and six-degree-offreedom industrial robots.Achieved the automatic recognition of different kinds of spraying workpiece poses,with the off-line programming technology which can realize flexible spraying.Firstly,we analyzed the needs of the project,and put forward a system solution to realize the recognition of spraying workpieces and pose estimation and analyzed the main structure and workflow of the system.Then we putted forward a calibration algorithm based on SVD theory.Then,we studied the effective preprocessing algorithms of point cloud extraction,filtering and streamlining,which deal with the interference of background,noise of the workpieces.Based on the feature extraction of point cloud,we established the offline feature model library for point cloud of CAD model.And proposed a method for workpiece recognition based on the ratio threshold of side length of bounding box and the method based on feature matching,which realized the workpiece recognition and pose estimation.After that,we studied the P-ICP registration algorithm for point cloud and get a good precise position of the workpiece.Finally,we set up the experimental system based on the hardware of industrial robots and wire-structured light sensors.And realized the system functions through the establish of communication.Then we use a laser tracker measurement system,measuring the position estimation accuracy through experiment and the result satisfies the accuracy requirement of the system accuracy,which verify the feasibility of our method.