Currently, the automation of loading of operating machines is done either using a completely flexible tool, the robot, or an extremely rigid mechanical auxiliary machine. The robot can position the wrist anywhere in the working volume with an arbitrary orientation, while an auxiliary machine, usually composed of vibrators equipped with trap paths, is designed to position and orient a single component, in a known position. This rigidity, combined with the need to shake the parts to move them by generating micro-bumps between them and the considerable bulk of such equipment, make this approach a suboptimal solution.
The implementation of a system capable of picking up randomly oriented parts by means of an anthropomorphic manipulator would overcome such problems and provide an extremely effective tool in the quest for production process optimization.
The earliest examples of direct bin picking date back to the late 1990s, several times applications were presented that effectively accomplished “Bin Picking” on simple parts. Despite these attempts and more than two decades later, Bin Picking remains one of the greatest challenges in the field of vision applied to robotics.