Verticality of hole is strictly required during the assembly of aerospace components. In order to ensure verticality when using automatic equipment such as robots to make holes

it is necessary to measure the normal direction of each machined hole in place. The existing normal measurement methods of the workpiece surface are difficult to meet the measurement requirements of the workpiece surface with large curvature. Through line laser scanning

point cloud data reflecting the detailed information of the local surface of the workpiece can be obtained. On this basis

a method for calculating the surface normal of the workpiece with large curvature is proposed. First studied the method of plane fitting to local point cloud data to obtain the surface normal through PCA (principal component analysis)

then selected some typical curvatures

generated point cloud data through simulation

and analyzed the change of normal fitting error law. The effectiveness of the above method is verified by experiments. The results show that when the selection range of point cloud is less than 12 mm

the normal fitting error of cylindrical test piece with diameter of 50 mm can be guaranteed to be less than 0.219°.