The Application Research Of Point Registration In The Robotic Belt Grinding System

Aeronautical blade and turbine blade have complicated profile，with high precisionrequirement，it mainly uses belt grinding as the finishing processing. However, the traditionalmanual belt grinding does great damage to worker’s healthy, and it is difficult for the workerto guarantee steady machining quality; As to the numerical control grinding machine, it hasthe disadvantages of low efficiency and high cost. In this paper, a flexible robotic beltgrinding system has been presented and applied in grinding processing. Because the path ofrobot has direct relationship with the workpiece coordinate system, but the relative positionbetween robot and workpiece is unknown, and it is difficult to calculate manually, so3-Dpoint cloud registration technology is used in robotic belt grinding system to obtain theworkpiece coordinate system and compensate the clamping deviation, keep constant materialremoval amount.The point-cloud to be matched are scanning point-cloud and discrete point-cloud of partmodel,because of the significant difference between their initial positions, PrincipalComponent Analysis(PCA) algorithm is used for coarse alignment, then improved IterativeClosest Point(ICP) algorithm is employed to fine registration, and the result can be convertedto the workpiece coordinate system.As the part model is not the exactly same as the workpiece to be grinded, in order tokeep constant material removal amount, it is needed to correct the path targets by means ofcomparing the scanning points with the corresponding points in the workpiece.There is little difference between the fixture location of different workpieces, i.e.clamping deviation, we use the improved ICP algorithm to get it directly.This paper take several kinds of turbine blades as experimental objects, the machinedaccuracy and surface quality can meet the actual machining requirements, which confirms theeffectiveness of the registration algorithm.