Machine Vision System Takes a Front Seat

Ruth A. Mendonsa

Lear Corp., a Canadian manufacturer of auto seat frames, needed a way to improve product quality and eliminate the problems in its production process. Misassembled auto seat frames are not only costly and detrimental to production rates, but also can cause damage to the welding equipment.
The seat frames come in two- and three-passenger configurations. Each contains many components whose order must be verified before the welding process. Machine vision was chosen as the most reliable method for verifying component placement.
Lear went to experts at Insight Integration of Lansing, Mich., who evaluated the process and determined that Lear needed to integrate a high-speed system capable of accurate component measurement into its existing manufacturing line.
The solution was VisionBlox, a set of machine vision inspection libraries from Integral Vision in Farmington Hills, Mich. VisionBlox has a set of machine vision software components used under Microsoft's Visual Basic to create programs with a Windows user interface.

Overcoming the obstacles
As with most complex machine vision applications, there were obstacles. The seat frames moved within the process in a palette/jig and were manually handled up to the point of inspection, making positioning an issue. And the large frame size demanded a field of view big enough to encompass all areas of inspection, so special selection and placement of cameras was necessary.
VisionBlox software starts and runs automatically with power up. In this application, the inspection tools are grouped by eight functions. Completion of all eight inspections of about 80 components takes 2 to 3 s. The assembly process has a throughput of about one seat frame every 30 s. A Microsoft Windows interface shows the operator a seat frame with defect indicators on the test locations. When a frame inspection fails, the failure is highlighted in red, indicating the component to check.
This system has satisfied both parties: Lear reports better overall quality and less scrap, and Insight Integration reported a reduction in the usual development time involved in customizing an application.