Abstract
Sand casting usually implies some flash or burr on the cast parts that are unevenly distributed and variations in size and mass. Furthermore, the cast part itself can also have significant geometric deviations caused by sand casting process variations. Effective automated deburring needs to be fast and accurate, adapted to individual parts’ burr size and the actual geometry of the cast part. This paper reports on an experimental study of adaptive one-of-a-kind automated pre-programming for robotic deburring using structured light 3D machine vision optical measurements for tool trajectory generation. Each individual cast part is scanned to create a 3D mesh that is used to generate the tool trajectory. The method is verified in an experimental setup with a robot manipulator with a pneumatic spindle and carbide cutting tool. This study can serve as an example on how vision sensors and advanced algorithms can be applied to achieve productive and adaptive manufacturing processes and systems.
