« Back To Vision & Imaging Industry Insights
AIA Logo

Member Since 1984


AIA - Advancing Vision + Imaging has transformed into the Association for Advancing Automation, the leading global automation trade association of the vision + imaging, robotics, motion control, and industrial AI industries.

Content Filed Under:

Automotive , Fabricated Metals , and Transportation Automotive , Fabricated Metals , and Transportation

Visual Inspection & Testing Visual Inspection & Testing

3D Vision Seamlessly Enables Perfect Welds

POSTED 12/16/2008  | By: Winn Hardin, Contributing Editor

Manufacturing industries use laser, MIG, and TIG welding to join metal parts in many durable goods, ranging from automobiles to washing machines. As metal and other commodity prices continue their steady rise, manufacturers are striving to restrain costs by using newer, lightweight materials that often depend on complex physical structures or new material properties to inject greater strength and stability into the final product. This means that welding operations can become more complex, more numerous, difficult to access, and require greater precision of the weld and tracking of the seams during the welding operation.

3D machine vision systems – laser-based triangulation systems in particular – are helping manufacturers meet these increasingly stringent welding requirements in a number of ways, including smaller sensors with larger fields of view that can access difficult areas, while simplifying system integration through sample equipment interface code to ease integration with PLCs and other common industrial automation equipment.

Saving the Automotive Market
In terms of the number of welds assisted by machine vision systems, usually in conjunction with a robot or as a quality inspection system after a manual weld, the automotive industry leads the manufacturing pack.

‘‘Automotive is the biggest market for machine vision in welding,’‘ explains Matt Huff, Technical Sales Representative at VITRONIC Machine Vision Ltd. (Louisville, Kentucky). ‘‘Using machine vision in automotive welding is fairly new to the U.S., but these systems have been in use in Europe for many years.’‘

According to Huff, traditional approaches to weld inspection place a human at the end of the production line to check each weld by eye. ‘‘And as we know, the human eye cannot detect the difference between 0.8 and 1 millimeter very well. So, not only are we doing it more accurately and repeatedly than manual inspection, but we do it at fixed cost.’‘

Huff adds that the current downturn in the automotive industry has actually produced more queries for machine vision in welding operations because of the potential to reduce labor, an item of contention among U.S. carmakers. ‘‘Most of the people we’re seeing are Tier 1 and Tier 2 suppliers. If they send a non-conforming part to an OEM, they get into a lot of trouble and have to have a third party inspect the part at a significantly higher cost. Machine vision is a great way to reduce that risk and avoid the problem.’‘

Emerging Opportunities
‘‘There are only a few other industries where weld quality is as important, such as heavy machinery from Caterpillar and John Deer,’‘ adds Huff. ‘‘The tolerances for heavy machinery are usually a little higher, but quality is still important.’‘ 

Flushing out potential welding markets for machine vision, LMI Technologies ( British Columbia, Canada) European Representative, Martin Sandén, adds, ‘‘Tube welding and metal sheet construction, larger objects with metal frames and vessels of all kinds are also important markets for machine vision systems to track and inspect welds.’‘

Welding applications present machine vision systems with some unique challenges. During the welding operation, sensors need high dynamic range to be able to see the seam through the intense light created by the weld head. Welding operations are messy, even by industrial standards, producing smoke and spatter that can damage sensitive machine vision equipment.

‘‘You can get around these challenges with smart mechanical systems, shutters, and other measures,’‘ explains LMI’s Sandén. ‘‘It’s not uncommon to use air and water cooling systems to protect the vision system. With our Seam Finder system that we’ve had in service for several years, we use shutters to close the sensor during welding, and we scan either before or after the welding operation, depending on whether we’re using the system to guide the weld head, or to inspect the weld after it’s done.’‘

Melding Vision and Welding
To make their systems more compatible with existing manufacturing operations, machine vision designers are using a number of ways to increase speed and throughput, while making it easier to image hard-to-access welds and connect to adjacent production equipment.

‘‘One of the challenges of machine vision in automotive welding is throughput,’‘ explains LMI’s Sandén. ‘‘Competing methods include contact sensors that every robot manufacturer has in their catalog, but these sensors need more maintenance than non-contact sensors like machine vision, and are slower and less accurate than laser-based machine vision. Cycle times are very critical to the automotive business. If a system is going to slow them down, they’d rather not use it.’‘

New sensor designs are helping to make machine vision welding systems more flexible as designers continually add new welds, or seek to hide welds in places consumers can’t easily see – such as the bottom of the car door between the inner and outer sheet metal ‘skin.’

‘‘We’ve been focused on slimming down the sensors to get them in more spaces,’‘ explains VITRONIC’s Huff. ‘‘We’re also looking at changing the view angles of how the camera looks at the laser line without losing too much spatial resolution. We’ve done some pretty tricky custom sensors for major automakers, but we’re looking at ways to scale that down into a standard product measuring just a few inches on a side.’‘

Simplifying integration is also a big differentiator for machine vision welding systems. ‘‘Cabling is more important than you would think,’‘ says Cognex Corporation’s (Natick, Massachusetts) Paul Vondrak. ‘‘Our cables are rated to 10 million flex cycles, which is about 10 times more than the industry standard. And in addition to making the system’s easier to build using point and click programming, Cognex also gives customers sample code for communicating with all the major robot and PLCs.’‘

Machine vision is continuing to evolve along side the need for industrial welding. Today’s systems do far more than tell a robot where the seam is or tell a customer if a seam falls within safe tolerances. Cutting-edge weld inspection systems tell customers what part was scanned, what welds were defective, the types of defects and where the defects lie along the weld. As vision systems continue to build more powerful features into simpler packages, the days of human welding for many applications may one day come to a ‘seamless’ end.