Point Cloud Based Localization Of Part Models In The Blank

With the gradual development of the manufacturing industry, for different workpiece positioning requirements and continuously improve the degree of flexibility. In this paper, based on parts of the point cloud model in the blank localization technology, has important theoretical significance and value.First, the analysis of the status quo based on the blank positioning of parts, as well as location algorithm provides an overview. The localization divided into the pre-location and precise location of the two steps. The localization is through an iterative algorithm to find the optimal transformation matrix, then its corresponding transform, to complete the localization.Second, this thesis studies pre-location of the blank based on the point cloud of part model. In this paper,there are two methods to complete the pre-location. In the one hand, the optimal rotation angle based on genetic algorithm, the minimum projected area of calculation of the point cloud corresponding to the angle of rotation angle and the distance of two point clouds to find the optimum amount of shift, by translation and rotation of the blank point cloud to complete the pre-location. In the other hand, by principal component analysis complete the pre-location. By calculating the covariance matrix of the two point clouds of eigenvalues and corresponding eigenvectors, sort the eigenvalues and align larger eigenvalues corresponding eigenvectors, and finally completed pre-location by manually rotating blank.Third, parabolic fitting and plane fitting to get the normal vector of the center of the blank grid cloud. By creating the grid of the grid and expansion, the growth method to find the external grid, build ANNKd-tree nearest neighbor search to find the nearest point of the parts of the point cloud and calculate the normal vector and distance. The remaining parts of the normal vector of the point cloud by Shepard interpolation. Adjust the normal vector direction to point outward.Fourth, the third step of the preparation, first just use genetic algorithms to precise positioning of the blank part. Second, the blank by pressure-based model for precise positioning, the use of advance and retreat method and golden section method to find the best transformation matrix to achieve the blank parts completely wrapped, and then based on genetic algorithm maximize the minimum distance between parts point clouds and blank point clouds to complete the precise localization. Finally, completed object-oriented localization of the blank in visual C++2003platform. The blank of hairclipper and blade localization as examples verified the effectiveness of the algorithm.