Research On Laser Detection Technology Of Heterogeneous Casting Surface Line

Due to plastic deformation,unreasonable casting process and other reasons,concave convex defects with different sizes and random distribution will be formed on the surface of heterogeneous castings.and die closing lines and pouring holes with different depth information will be left at the end of casting.Direct grinding will not only affect the processing efficiency,but also accelerate the wear of grinding equipment.For a long time,heterogeneous castings often rely on manual appearance defect detection and dimension measurement before processing,which is time-consuming and inefficient.With the rapid development of computer,numerical control,sensor technology and robot technology,the digital measurement of complex surface has made a major breakthrough.In this paper,based on the principle of laser triangulation,a detection system of heterogeneous parts surface based on line laser is proposed.The hardware system and software platform are built to identify the defect information and machining allowance and guide the subsequent grinding.Focusing on the calibration accuracy,point cloud processing,defect detection and other issues,the corresponding research is carried out,and the detection of bathroom faucet is completed.(1)Robot hand eye calibration based on standard ball is studied.By analyzing the principle of laser triangulation,the eye to hand calibration mathematical model is established.Aiming at the low calibration accuracy and poor robustness of the calibration system,a fitting algorithm based on Tukey weight function reweighting is proposed to optimize the spherical center fitting.At the same time,the robot motion error introduced in the calculation process is optimized by weighted least square,From the data source and data processing,the calibration accuracy is improved and the robustness of the calibration system is enhanced.(2)The point cloud denoising and reduction method is studied.Through the analysis of point cloud data types,this paper preprocesses the line scan point cloud data.For the discrete noise and mixed noise encountered in the process of point cloud collection,statistical denoising and mean filtering are used to denoise.In addition,for the redundancy of line scan point cloud data,the point cloud characteristics are analyzed,and the point cloud simplification is completed by using the neighborhood based voxel grid method.(3)The point cloud registration method is studied.According to the practical needs of the project,through the design of coordinate transformation scheme,the point clouds of standard parts and blank parts are unified into the same coordinate system,so that the translation and rotation dislocation between the point clouds are reduced without relying on the algorithm,and the coarse registration of the point clouds is completed.Finally,the precise registration of the point cloud data is completed through the GICP algorithm.The calculation of dimension error between standard part and blank part is realized.(4)The laser rapid detection system of robot is studied.The hardware of the system is selected,the laser detection software is redeveloped,and the overall workflow is designed.A certain type of bathroom faucet is selected as the research object.The scanning data of workpiece surface is collected and processed by self-developed software to obtain the surface dimension information,and then the defect information and machining allowance information are obtained.The line laser rapid detection is completed,which meets the engineering needs.