Fast Method for Triangulation of Point Clouds Obtained from Computer Vision Systems of Industrial Robotic Complexes

Abstract

In summary, the paper proposed an algorithm for the fast triangulation of point clouds obtained using computer vision systems based on profile-type laser scanner. The proposed method uses the features of the formed point clouds. The incoming at each scanning step data from laser scanners is formed as a point cloud representing the profile of the surface falling into the scanning plane. These profiles are located one behind the other with a given step and allow to easily detect points for constructing triangles that form a polygonal model. This fact makes it possible to reduce the computational complexity of converting these points to polygonal models tenfold compared with traditional triangulation algorithms. Because of polygonal model usage, it is possible to correctly set the working tools movement trajectory and their desired orientation in the process of movement.