Robotic 3D Grinding Path Planning Based On Point Cloud Data Processing

With the rapid development of intelligence and automation,robotic grinding is gaining more applications in manufacturing processes such as casting grinding,mold polishing,blades machining,deburring,sanitary ware polishing and so on.Compared with the traditional manual grinding,robotic grinding relieves hand grinders from their laborious work and bad work environment,but also achieves efficient,high quality grinding surface.In the robotic grinding process,the path planning is a very important part that has a direct impact on the grinding quality.Generally,the robotic grinding path planning is realized by teaching method or off-line programming based on the CAD model.However,in the process of the blades' repairing,the original CAD model cannot be used for grinding path planning because of the deformation and distortion after working in the high temperature and high pressure conditions.Therefore,how to optimize the grinding path rationally before the robot grinding process,and use the point cloud data processing to achieve the grinding workpiece model and grinding process simulation are pressing research problems to be addressed in this thesis work.This thesis presents a robot grinding system based on 3D measurement to realize the grinding of aeronautical parts such as blades.This system uses a Keyence laser scanner to obtain the point cloud data of the workpiece.After the point cloud data pre-processing,the grinding contact points and attitude during the blade belt grinding process are planned.During the planning of the contact points in the blades grinding,an octree-based point cloud slicing algorithm is proposed to find the intersection of the plane and the point cloud data.In the octree division,the algorithm does not divide the leaf node until only with one point,but selects a partition depth.The speed of point cloud slicing algorithm is increased,while avoiding the disadvantage of the octree dividing efficiency on high density point cloud data.Based on the curvature grinding,a contact points discretizing method is proposed by comparing the equal step method with the Douglas-Peucker algorithm.According to the characteristic of the curved surface trajectory,the linear equation is selected as the parametric equation for pre-discretization.Then the Douglas-Peucker algorithm is used to insert points to meet the precision requirement.The results show that the proposed method can meet the requirements of curvature grinding.The process to plan the attitude during robotic grinding can be regarded as to estimate the normal vectors of point cloud data on the contact points.Based on the comparison of existing point cloud data estimation algorithms,a normal estimation method based on KD tree neighborhood fitting is adopted.Then,the normal vectors are filtered to ensure that the normal between adjacent contact can transfer smoothly and that the smooth robotic belt grinding movement can be achieved.The research laid a theoretical foundation for robot 3D grinding path planning,and the direct application of point cloud data processing technology to the robot grinding process can not only achieve the grinding of the parts without CAD model or poor consistency,but also advance the online or real-time adjustment based on the measurement point cloud data during the robotic grinding process.