Controlling the welding process with machine vision

There’s nothing particularly new about using vision to locate edges for a laser weld – here’s an example from Southwest Research to prove my point – but all the applications I’ve seen have been two dimensional. By that I mean the camera is used to locate the edges in X and Y directions and guides the laser accordingly. But no one ever seems to consider the third or Z dimension – the depth of the weld pool.

This is a big deal for car makers because they go to a lot of trouble to hide unsightly welds from view. “Wouldn’t it be nice,” they often say (or so I imagine,) “If we could join two sheets and not have the weld show?” (The problem of course is that the laser creates a weld pool that melts right through both upper and lower sheet.)

Well it turns out that clever people at the Fraunhofer (where else?) have also been thinking about this, and they came up with a way to monitor the weld pool depth. “Perfect welds for car bodies” gives an overview of a thermal imaging system that “calculates the penetration depth into the lower sheet. A software program then adapts the output of the laser to the specific requirements.” And in this way the perfect weld is created.

Details of the technology are noticeably absent from the Fraunhofer’s press release, although they do mention the need for extremely rapid image analysis. Apparently, to deal with this they are utilizing “A tiny processor … integrated in [to] each pixel.”

I’m not sure I’m buying that, but I guess we’ll have to wait until the patent application is published to learn more.