Automated inspection of mechanical components by optical systems

In the mechanical field agile production requires rapid and flexible shop-floor control techniques in order to verify the manufacturing operations effectiveness. To accomplish this goal, 3D scanning systems are evermore used for quality control purposes. Generally, 3D CAD (Computer Aided Design) product models can be used as reference in order to manage the verification process. The geometrical information represented in the 3D CAD model can be used in order to virtually plan, simulate and pilot the inspection process. The present work targets the development of view planning algorithms in order to support the automatic inspection of 3D shapes, including dimensional and geometrical tolerances. Algorithms have been implemented in a prototypal software system that has been experimented as off-line application to provide inputs to a multi-axis Degree of Freedom (DoF) robot arm mounting an optical 3D scanner. Two test cases from automotive and die casting fields are presented. They show the computation of acquisition poses in a suitable sequence and correspondence to the experimental data.