System integrators use specialized skills for complex vision applications
The commoditization of machine vision has given customers the option to select off-the-shelf solutions for their applications. Take smart cameras, for instance, which embed lighting, software, and input/output (I/O) interfaces — and eliminate the need for system integration. The only problem? These plug-and-play systems don’t play so well when the vision application is difficult.
“When you get into more complex inspections or really high-speed applications that require advanced tools and a specialized skill set, the system integrator becomes invaluable,” said Darcy Bachert, CEO of Prolucid Technologies Inc.
System integrators across the industry are fielding more requests to develop complete and detailed systems. Pyramid Imaging cites two primary reasons: More companies are recognizing the importance of machine vision, and small and midsized businesses need additional help from system integrators than their larger counterparts. “Many of them view us as their machine vision department,” said Dr. Rex Lee, president of Pyramid Imaging.
The increasing demand also comes from users who expect speed and precision in machine vision and want to take advantage of the latest technological advances. “It used to be that you took one picture of a part going down a conveyor and decided whether it is good or bad,” said Brian Durand, president of i4 Solutions. “Now we might be taking many images of that part, perhaps under different lighting conditions or at different angles, gathering more types of data about that item, and applying technologies like 3D imaging and machine learning.”
System integrators need to be adaptable enough to consider a variety of requests from customers, find the right partners if they are unable to fulfill those requests, and let users know if their vision project isn’t feasible. i4 Solutions, for example, is seeing more requests for projects that go beyond standard vision systems — think vision-guided robotics or packaging inspection — to more nontraditional solutions, such as a handheld device capturing aerospace-related information.
“If you have something that is portable, for example, the computational resources are far less,” Durand said. “It might require battery power or wireless connectivity, as opposed to the regular factory floor, where you have an Ethernet cable.”
When customers need to justify their machine vision investment, they rely heavily on — and demand more from — their system integration partners. But integrators do what they always do: Follow best practices to thoroughly examine a vision application. That comprises a proof-of-concept or R&D study, which may include testing samples in the integrator’s in-house vision lab.
“At this stage, we discover any serious engineering problems or limitations of the vision system,” Lee said. “It helps mitigate the risk of any unknowns.”
Additionally, this phase “makes it well-known to the customer that we can solve the problem,” Bachert said. “But we also know how much it’s going to cost, how long it’s going to take, and what the approach will be.”
Once feasibility is determined, system integrators conduct more testing as they build the system and train the user to operate the final product. “We have to be absolutely sure that the system is ready to go before we deliver it,” Durand said.
Before they put together a vision system, however, system integrators first act as teachers. “It’s our job to educate them and set the expectation of what is and isn’t capable with machine vision, whether they want their system to do one particular task or to easily adapt to many different types of inspection tasks,” Lee said.
Because customers are always asking about the latest technologies, it’s imperative system integrators stay ahead of the curve. Deep learning and 3-D imaging remain hot topics, but Bachert sees more traction in the cloud and Big Data.
“One of the big use cases we see is customers trying to build up larger classified data sets that include both good and defect images,” he said. “Once they have that dataset built up, you can make the algorithms more intelligent to drive down false rejects or improve the accuracy of detecting defects.”
The demand for big data prompted Prolucid to develop a dedicated cloud team about six years ago, while a separate integration team works exclusively on vision system development.
“We hire for both teams separately, because finding someone with both cloud and integration experience would be impossible,” Bachert said.
While smart cameras can solve many applications without integrators, complex projects that require design and other specialized skills are coming to market all the time. To be successful, machine vision system integrators must thoroughly understand — and stay one step ahead of — the newest technologies while adapting to ever-evolving realities in the industry.
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, firstname.lastname@example.org.