Machine vision comes of age
The first IEEE conferences that combined the words automation and comes of age were held in the 1970s. But machine vision industry insiders may remember 2017 as the year the world woke up and embraced the benefits from machines that can see as well as do.
"It has been a very strong year for growth," said John Merva, vice president, North America, at Gardasoft LLC, supplier of LED lights and controllers. "We had some very large projects last year, and yet we still managed to make that up and have a stellar year." Strong growth hasn’t been limited to the Northeast Corridor; it has reached every corner of North America. "Quite honestly, the US has been behind the curve of other industrial countries [regarding automation adoption], said Dr. Rex Lee, CEO and president of Pyramid Imaging, an automation system integrator and distributor.
"Europe and Japan have been leading in the adoption of manufacturing automation, while the automotive industry has led in the US. But now we’re seeing automation and robotics really grow in every manufacturing industry. Machine vision in particular has reached a price point that essentially [makes it] ubiquitous. Imaging sensors have become extremely affordable. And, the performance has increased to allow people to start implementing imaging sensors in so many different devices.
"Add to that the awareness in the engineering community, which has increased dramatically, on what imaging sensors can do to help them automate their solution, and it’s driving machine vision far beyond the traditional factory applications," Lee said. "It’s putting automation in medical devices, in agriculture. Honestly, I can’t think of a single industry that’s not using imaging sensors for some intelligent analytics."
Lee, who has extensive experience serving government, defense, aerospace, and related industries, says that as expected in today’s technology marketplace, North America’s commercial sectors are outperforming government, which still suffers from periodic budget uncertainty.
In addition to greater visibility for machine vision across the industrial sector and lower costs per pixel, Teledyne Dalsa’s Ghislain Beaupré, vice president of R&D and operations, noted the cost of processing power also continues to drop.
"Today we have enough processing power for just a few dollars to do some very interesting applications," he said. "The biggest challenge now is to keep adding qualified people to handle the growth. It’s a great situation to be in," Merva said, echoing comments shared by Lee and virtually every other AIA member company—an irony considering recent discussions about the effect of automation on job markets.
"The recent growth in vision and imaging technology is in part a result of a larger trend toward comprehensive automation solutions and connected devices," said Alex Shikany, director of market analysis at the Association for Advancing Automation (A3). "As these trends continue to drive investments by end customers, demand for machine vision and imaging technologies, such as cameras, lighting, optics, imaging boards, smart cameras, and image sensors, will continue to rise. At the same time, we’re seeing continued pressure in this market for embedded solutions that are smaller, cheaper, and more capable, which can be used in a variety of form factors and OEM products. This will challenge our industry to continue innovating, and AIA members have shown that they are up to the task. We’ve seen many new products and solutions emerge already which are aimed at solving these complex challenges."
What drives tomorrow’s growth?
In addition to cheaper pixels and transistors, machine vision standards and software will also help propel imaging technology to every corner of the electronics universe.
For example, deep learning is more than a buzz word, according to Beaupré. "For deep learning, you need a lot of data, and imaging does exactly that."
Deep learning is a subset of, or precursor to, artificial intelligence. After being shown a sufficient number of good and bad images of the application under development, deep learning software can automatically program a machine vision system in minutes—a process that would otherwise take many months.
"You let the computer create the algorithm as opposed to a person," Beaupré said. "Applications that were too difficult to do, or required too much investment, will soon be possible, which will further open the door for machine vision solutions." Gardasoft’s Merva adds that deep learning won’t just affect algorithm performance but will also help control cameras and lighting systems. There is no limit to the affect deep learning will have on machine vision, Merva said.
Some standards are helping customers tackle very large applications for less investment. For example, IEEE 1588 provides a highly accurate network clock for synchronizing multiple systems, including smart cameras. Other standards are helping customers choose the best solution for their application. For instance, EMVA 1288 standardizes camera specifications and materials.
Today, machine vision companies are simplifying every aspect of using their products, from ordering to implementation. As machine vision crests a new wave of visibility across mainstream engineering communities, the exceptional growth of 2017 may one day be seen as a threshold moment—the year when everything changed and the only recalibration necessary might be what the industry expects is possible tomorrow.
Winn Hardin is contributing editor for AIA. This article originally appeared in Vision Online. AIA is a part of the Association for Advancing Automation (A3), a CFE Media content partner. Edited by Chris Vavra, production editor, CFE Media, email@example.com.