Abstract: Dimensional measurement of the subassemblies at the beginning of the assembly line is a very crucial
process for the eyeglasses industry, since even small manufacturing errors of the components can lead to
very visible defects on the final product. For this reason, all subcomponents of the eyeglass are verified
before beginning the assembly process either with a 100 percent inspection or on a statistical basis. Inspection
is usually performed by human operators, with high costs and a degree of repeatability which is
not always satisfactory.
This paper presents a novel on-line measuring system for dimensional verification of small metallic
subassemblies for the eyeglasses industry. The machine vision system proposed, which was designed to
be used at the beginning of the assembly line, could also be employed in the Statistical Process Control
(SCP) by the manufacturer of the subassemblies.
The automated system proposed is based on artificial vision, and exploits two CCD cameras and an
anthropomorphic robot to inspect and manipulate the subcomponents of the eyeglass. Each component
is recognized by the first camera in a quite large workspace, picked up by the robot and placed in the
small vision field of the second camera which performs the measurement process. Finally, the part is
palletized by the robot. The system can be easily taught by the operator by simply placing the template
object in the vision field of the measurement camera (for dimensional data acquisition) and hence by
instructing the robot via the Teaching Control Pendant within the vision field of the first camera (for
pick-up transformation acquisition).
The major problem we dealt with is that the shape and dimensions of the subassemblies can vary in
a quite wide range, but different positioning of the same component can look very similar one to another.
For this reason, a specific shape recognition procedure was developed. In the paper, the whole system is
presented together with first experimental lab results.