Latest question:
How do you think the new GigE standards will influence the machine vision industry?
Respond or ask your question now!





By Dr. Josep Forest, AQSENSE SL
Side car-body parts have to be assembled with the roof and chassis, or doorframes must fit their corresponding bodies up to a certain and specific accuracy. These are critical issues in the automobile manufacturing process because the parts simply must fit together very well if we don’t want to drive vehicles that are noisy, shaking with vibrations and have soaking-wet interiors from leaks when it rains.
Until now, the problem has been addressed by making point measurements—as far back as 1986 a complex Automatix system was installed at Cowley, England, using 62 laser cameras to carry out 96 direct and computed checks on the body shell. By contrast, it now is possible to literally check millions of points with just a few scanners while still making use of laser triangulation. Usually these kinds of parts are manufactured through stamping, a process involving powerful mold press systems, at a pace of several parts per minute, a part meaning, for example, a whole side body, a door or a whole chassis.
Even when metal sheets for car body manufacturing are subject to tight quality inspection to guarantee evenness and strength to a certain tolerance, and in spite of the stamping process being of high quality, discrepancies between the produced part and the designed part still may appear. Only 100 percent of parts inspection can guarantee the dimensional tolerances to obtain the correct fits (see Figure 1).
Metrology tools can be used to inspect one single part out of several, involving the use of complex tooling hardware and software, which are perfect tools for very complete dimensional analysis, including full part scanning, CAD comparison, gauging or automatic reports generation. However, this process is very time-consuming, requires highly skilled staff to operate the systems properly, and is definitely not oriented for in-line 100 percent of parts inspection tasks.
Commercially available machine vision tools can be used to solve a number of applications at the production line, mostly oriented to 2D image processing. However, there is a lack of 3D processing software for machine vision applications.