Vision Systems Design celebrates its 25th anniversary in 2021. What are some of the most interesting and notable machine vision advancements you’ve seen since you entered the space?
I first started in this space fresh out of graduate school in 1987 when I was hired as a software engineer at Datacube by the late Stanley Karandanis, a very influential player in our industry. In these 34 years, I have witnessed and been a part of several major and notable advancements in our industry.
State-of-the-art in vision in the 1980s was the beginning of the transition to move from the use of broadcast analog video cameras (e.g. NTSC/EIA-RS170, PAL/CCIR) in vision applications to analog progressive scan area cameras that were capable of being more precisely triggered and had programmable exposure control to work with industrial lighting. This really helped enable the expansion of vision into industrial factory automation applications. In those days, there were no monolithic analog to digital (A/D) converter chips available, so being an electrical engineer meant you had to have a solid understanding of analog design and programmable analog image acquisition meant complex analog designs that consumed a lot of board real estate.
In the early 1990s, a rapid expansion of industrial cameras occurred, including line scan, and the first digital cameras were becoming in vogue. In those days, host computers such as PCs did not have nearly enough processing horsepower to handle the image processing tasks, so this meant lots of purpose-built image processing hardware based on programmable logic (PALs and early FPGAs) and even expensive custom silicon or ASIC designs. Through the 1990s, when Vision Systems Design first entered the market, we were undergoing a rapid transition of vision systems occupying racks of space in high-end computers to desktop PCs. This was a rapid transition as Moore’s Law finally got PCs to the level of processing power that matched, or even exceeded, the performance of custom-built hardware.
Around the new millennium, this transition was occurring so quickly that many former leading companies in our industry could not keep up with the pace of the PCs. This was when the vision industry was able to “ride the wave” of consumer-driven technology advantages, and the era of standardization entered the vision space with Firewire and GigE Vision being of significant note. This ushered in the rapid expansion of camera offerings and the race to offer higher resolutions at higher frame rates for an ever-increasing demand for more pixels in new application areas.
In the 1980s, we were talking about cameras generating 7 to 10 MBps of data that had to be acquired and processed. In the 2000s, this started to reach 100 MB/s, which was possible with highly optimized GigE Vision cameras and device drivers.
Fast forward to today, and from where I sit in my role at Euresys, even 1 GBps of data feels like the “lower end” of the types of applications we are working on with our customers. With the latest interface standards including CoaXPress, CoaXPress-over-Fiber, Camera Link HS, and 10G/25G/100G GigE Vision, it is not unusual to consider image bandwidths in the 7 to 10 GBps range.
Looking toward the future, what sort of advancements are you most looking forward to?
From my perspective, what continue to drive our industry, quite simply, are technologies that are driven by consumer demand. Think of the consumer broadcast and Internet bandwidth advantages, 4K (~8-MPixel) flat panels and broadcast content of 120 Hz, and now 8K pulling our industry along with them as the needs for automation of production and inspection of these consumer devices, including the mobile phone in your pocket, increase. The components and technologies developed for high resolution, high frame rate broadcast digital television, and the chipsets and technology for huge increases in network devices are all directly applicable to and utilized by the vision industry. Further, the demand for gaming, graphics, and artificial intelligence coming from the consumer industry is making available image processors like GPUs, embedded FPGAs with multi-core ARM processors, or specialized embedded computer platforms that provide the processing horsepower necessary to make the application and use of deep learning and neural net execution possible in real time, on the fly. Remember the first 16 MPixel sensor? Today, it is more and more common to work with 45, 65, and even 150 MPixel sensors that have become mainstream and affordable.
Do you have any notable Vision Systems Design-related stories to share from over the years?
Vision Systems Design has been part of the fabric of our industry for quite some time now. Most notable stories for me always have the late Editor-in-Chief Andy Wilson involved in some way. Andy had both a broad and impressively deep understanding of not just the technologies and components side of the industry, but also the application use cases that he experienced firsthand and wrote about. Reading Andy’s monthly editorial was always the first article I read in every new issue—he had an especially effective way of integrating his quick wit and humor into the general theme and content of every issue. Knowing Andy in person and working with him over the years was a great honor.
What are you most excited about at your company right now?
As we are a developer and supplier of key components and enabling technologies, such as our image acquisition devices (frame grabbers), video processors, vision software, and IP Cores, we regularly come in contact with applications in emerging markets. Learning about these new application spaces feels like we are at the forefront of the vision industry. Seeing how rapidly stadium imaging is evolving to include consumer-driven and customizable viewing experiences, the use of advanced scene and play analysis, and even real time capture of all the play action in 3D is really inspiring. The push for autonomous vehicles, transport, logistics, and medical procedures regularly presents us with really new and exciting application areas to be involved in. What I appreciate most is that we position ourselves as a development and support partner for our customers rather than just sellers of technologies. This gives us great insights into the latest trends and requirements that allow us to advance our product offerings.
What is your latest product, and for what reason was it developed?
Speaking of advancing our product offerings, we are particularly proud of the fact that we have made possible for the entire industry the next level of high-bandwidth, yet simple and real-time critically reliable transport of camera data to host PCs with the development of CoaXPress-over-Fiber. While we developed this initially for several customer opportunities that were pushing for this advancement, we have developed it in a way that the entire industry can benefit from our work. A major milestone of that was the adoption of CoaXPress-over-Fiber by the JIIA-led CoaXPress standard committee and the now public release of this as an official advancement of the standard.