Machine Vision Moves Forward
LEARN MORE ABOUT TODAY’S MACHINE VISION.
If you’re interested in self-driving cars, drones or robotics, then you will understand why machine vision continues to grow. These futuristic technologies are a few areas where machine vision works—and the traditional application areas continue to do well.
According to research from the AIA trade group, the machine vision market had its highest first quarter total in history in the first quarter of 2015. Sales of machine vision systems and components in North America grew 22%.
The machine vision systems category saw a year over year increase of 24%.
In the first quarter, smart cameras expanded by 23%, while application specific machine vision systems increased 24%.
Machine vision components had a strong quarter with 11% growth over the first quarter of 2014. The leading product categories within machine vision components in terms of growth were lighting (28%), cameras (11%), and software (8%).
While sales are on the rise, the technology itself continues to improve. From CMOS sensors to 3D imaging, learn what today’s trending topics are in machine vision.
Machine Vision Today
“It’s an interesting and growing market today,” says Alex Shikany, AIA’s director of market analysis. “There are becoming a lot more uses for vision and imaging technologies outside of traditional machine vision applications.”
In addition to factory automation, a traditional and still important machine vision market, Shikany says life science is a growing area, along with automotive and vision-guided robotics—“We’re seeing that really take off,” Shikany says—and even UAV, or unmanned aerial vehicle applications. The range of products continues to grow, with more capable components and systems, along with exceedingly large resolution cameras offering high speeds and frames per second.
While the manufacturing space continues to adopt machine vision technology, experts say that it is only getting better.
“Efficiency and quality are two huge value propositions of machine vision,” Shikany says. “Another one is speed. You can produce much more when your system has eyes. Computers can process at a very fast rate, and increase your throughput, which also leads to more sales.
“In today’s manufacturing climate, more small and medium-sized companies are going with machine vision,” he continues. “One of the main things is that it is getting more affordable, in component categories as well as systems categories.”
As the industry continues to grow and technologies continue to improve, experts say it is more of an evolving change rather than an overnight one.
“We’re seeing rapid movement in a variety of areas but not so much a big ‘aha’ moment,” says David Dechow, FANUC America Corp. material handling segment staff engineer - intelligent robotics/machine vision. Dechow says there has been a trend in integrated devices, 3D sensors, and new camera offerings. “It’s really been jumping in the last six, eight and twelve months,” he says. Cameras are adopting USB3 and GigE vision technologies, and there has also been a continuing move toward CMOS sensors, with higher resolution and faster speeds, he says. This improved CMOS technology is allowing cameras to develop better capabilities. And even in mature products such as light sources and optics, Dechow says improvements are happening.
“Three-dimensional imaging is certainly being embraced. I think it’s going to get more mature, usable, and more applicable to a wider variety of applications.”
Shikany adds that the shift from analog to digital is an ongoing trend. In the past, companies may have used very inexpensive analog cameras. “Now almost everything is digital,” Shikany says. “Now, in two to five years in my estimation, even on the low-end everything will be digital.”
Ben Dawson, director of strategic development at Teledyne DALSA Industrial Products, predicts more innovations in the industry as well.
“Smart cameras are going to get smarter,” Dawson says. New smart cameras may not look much different, but they can contain bigger sensors, such as a new camera with a 5 megapixel sensor, and faster processors. “Larger sensors are a dual-edged sword,” Dawson says. “On one hand, they are better at finding defects. But they also require a better lens and more computational power.”
On the software side, Dawson says he follows computer vision—machine vision’s academic cousin. “We’re always looking at what’s coming out of the universities. A lot of algorithms are interesting, but their performance in the machine vision world is just too slow. A job I’m working on right now runs at 20 complex parts a second. I don’t have a lot of time to hang around.”
Research in computer vision includes smarter algorithms using machine learning. Big data, machine learning, neural nets (a form of machine learning) are all buzzwords in the field, Dawson says. The hope is that a machine vision system could be trained by giving it a large number of sample images and then “spit out a solution.”
And just as the technology has improved, so has the level of awareness for those using it.
“The good news is that manufacturing engineers and QC engineers are more accepting and more knowledgeable about machine vision,” Dawson says. “Some quality control engineers used to not want machine vision on the factory floor, they were afraid for their jobs.”
Today, he says people are more accepting of the technology, but sometimes expect too much. Because humans are so good at seeing things, he says that vision tasks can sometimes seem easier than they actually are.
Dechow agrees that machine vision still has challenges. “I don’t think that anything we’ve done to date with machine vision has completely solved the laws of physics when it comes to lighting, optics, part presentation,” Dechow says. “As components, algorithms get better and better, people may forget that we have to competently integrate lights and optics. I think the improved components and controls and wider variety of light sources and components does start to make it easier to use, but it doesn’t completely overcome the fact that good conscious design of components is required.”
As things get easier and easier to use, “maybe people take a little less care in what they are putting together,” Dechow says. People may think machine vision has “rounded the corner somehow and will work automatically.”
The Next Big Thing
The next big thing to watch, according to Shikany, is “how machine vision fits in with the factory of the future or the Industry of Things, industry 4.0. All that mashed together, making operations in your factories smarter.”
For those who aren’t sure about machine vision, Shikany says, “Stay tuned, there’s a lot coming and it’s only get to get better from here.”
“Pretty much any interesting advance, mechatronics, robotics, manufacturing, includes some form of sensor or machine vision,” says Shikany, citing the self-driving car, the factory of the future, smarter and more efficient homes.
The 3D imaging realm seems to be garnering the most interest from end users, according to Dechow. He says the continued improvement of lower level smart sensors is something to get excited about. These may be appealing for their ability to do simpler applications with less programming.
In the future, Dechow says he anticipates complex solutions being packaged in such a way that it’s easy for a customer to use. A whole system would be configured to solve a difficult task such as bin picking.
“The customer base and the end user are becoming more aware of the future needs of automation,” Dechow says. “Machine vision is such an enabling technology to facilitate the kinds of automation that’s going on on the plant floor nowadays.”
“A big part of the whole discussion of where machine vision is going is tightly tied into the needs of the automation industry,” Dechow says. “The systems that are being built now, enormously flexible types of automation, are what we need in the industry and the economy. In general, it makes the world successful, in putting together things that we need.”