Unfortunately, there is no simple secret; no Rosetta Stone that can be used to translate every desired inspection task into a smoothly running online machine vision application. When it comes to machine vision systems integration-the task of designing the correct lighting, optics, cameras, part placement, inspection algorithms, automation and then making it all work together to deliver the required results-the integrator must have 1) a thorough knowledge of the technology, and 2) plain old experience.
In the case of the technology question, the Automated Imaging Association (AIA, Ann Arbor, MI) has taken a major step forward in offering advanced Certified Machine Vision Professional (CVP) coursework during the 2011 Automate conference sessions in mid-March. These sessions, and the accompanying test, were a follow-up to the basic CVP offering introduced in 2010, which also was available in 2011.
This is the first time that a third-party association has undertaken certification of individuals who integrate machine vision technology-a process of certification that is quite common in other professions. The course material is relevant and, in some cases, downright difficult. Overall, the tests appear to be a reasonable indicator of technical knowledge. It has been a major effort, and the AIA should be commended for this huge contribution to the industry.
The program is not totally without some bugs, however. The principal goal of certification of a machine vision professional should be to provide an indication of that individual’s capability to utilize the technology in real-world integration and application. To that end, some of the CVP material might be too esoteric and textbook to be a real-world indicator, and those are the things that the AIA likely will be working out in months to come.
Ultimately, though, one must return to the issue of plain old experience in machine vision integration. No test in the world can fully evaluate real-world experience. It’s like an aspiring pilot who has taken all of the coursework in ground school and successfully passed the written exams. Despite having proved extensive fundamental technical knowledge, no one would want that pilot behind the controls of a 757 until he had accumulated literally thousands of hours of actual experience.
To be sure, many of the basic and advanced CVP individuals absolutely do have years of experience in machine vision, and it is that experience alone that fully indicates the ability to execute the successful integration of a machine vision application.
What's Next in Machine Vision?
Automate 2011 featured many new components, and in particular there were many new offerings in digital camera technology. There has been significant progress in the machine vision industry in the area of digital camera interconnect and data communication. A few years ago, the GigEVision camera interface standard and the GenICam standard for device description made direct digital interface of cameras to processors a realistic and viable for plant floor automation.CameraLink and now CameraLinkHS provide ultra-fast digital interfacing of cameras to a dedicated frame grabber and remain the gold standard in speed. A new interface standard, which is already commercially available, CoaXPress, may provide features that overcome some of the physical and speed limitations of previous interfaces.
All of these advances are for cameras, which will interface with centralized processors, specifically PCs. This technology is widely used in OEM solutions and assembly-specific machine vision (ASMV) devices. Is there a subtle movement in the industry back toward PC-based machine vision for general-purpose applications?
Certainly, smart camera architecture will not be displaced as the leading technology in general-purpose machine vision. However, with advances in interconnectivity, falling component price points and improving ease-of-use, the centralized PC-based solution may continue to grow as a real and attractive option for industrial machine vision integration.
