Chris Yates is the Managing Partner of Vision Ventures, a mergers & acquisitions (M&A) firm that specializes in the machine vision space. Concurrently he serves as the President of the European Machine Vision Association. He is not only a very busy person but also someone who has a uniquely informed presence in the market, particularly as it impacts manufacturing, and how it affects allied technologies such as robotics. He generously shared his time and thoughts about the present and future. In this second and final installment of the interview, we talk about combining machine vision and robotics; new navigation methods for mobile robots; the ability to have solutions that meet all three criteria of good, fast, and cheap; the balance of technology absorption and innovation releases via startup or incubator; and practical advice for fabricators ready to begin (or continue) technology integration.
FWM: One of the places we now see machine vision technology is attached to robotics. A model is evolving where you have an autonomous mobile robot or AMR, and you have a robot arm on top of that, and on top of that you have a machine vision system that shows the robot how to best sort the parts that it’s doing, or whatever the task is. I think that combination of technologies is set for spectacular growth, but you would be a better judge of that. To me, it’s an explosion waiting to happen.
Yates: I think robotics and vision have a natural and perfect synergy. It’s an automation of a mechanical system that is some way can duplicate the vision of the cameras. The camera, software, and processing provide eyes for the system. It’s a natural thing. Robots often need some level of fine guidance, unless it’s on a very well-defined path that is the same every time. There’s almost always some need for some kind of vision-based guidance there, whether it’s to pick up the particular part, and/or to identify exactly where it is. You’ll see a tremendous amount of vision around robotic installations, wherever they are.
But there is also the other side, which is robot-guided inspections, because of course you can put a camera on the end of a robot arm as the end effector tool, and then use the robot to move it around and do very flexible inspections.
I do agree with your assessment, AMRs have a tremendous requirement for vision, they require it for guidance. Navigation and safety are very important, and this is built into the AMR. As you say, there are robots also placed on top of this to do, for instance, pick operations in an e-commerce warehouse. They identify the barcode, pick the right box, go into the box and find the right article. So yes, it’s a perfect synergy, and it’s why you’ll see a lot of that at the trade shows like Automate in Detroit and Automatica in Munich. Vision companies, robotic companies, and automation companies are working on this together.
FWM: It’s unstoppable, seemingly. To explore a point about the AMRs, most of them are using LIDAR to get around the factory floor. I’m meeting with a company that is doing this with ultrasound—I never thought of that as an option.
Yates: Yes. This is a Norwegian company.
FWM: Absolutely right, Sonair [based in Oslo, www.sonair.com].
Yates: There is no reason to not use this technology. I do not know this product in great detail, but I know that there have been companies over the years that explored acoustic imaging. This is non-visible imaging, and it’s very interesting. In this case we are not using light at all. There are quite a few startups and early stage companies in this area. The company you mention is using ultrasound for guidance. It comes down to producing the data that is required to navigate. I could imagine there would be a cost benefit, and I am for all of it, the more sensing technologies have that work to provide guidance is good. All have different tradeoffs in terms of what causes problems in their “seeing.” For example, in visible wavelengths, dust and fog can cause problems. However, sonar or other acoustic technologies have their own areas where it will work fine and other areas where it will be less than optimal.
FWM: Having just welcomed our third grandchild, we can say without hesitation that ultrasound imaging has come a long way.
Yates: <laugh>. I know! I’d actually mentioned that to my partner about our daughter that I don’t even remember looking at it because I couldn’t make out what the picture was. And it’s very different now, as you say. It’s an interesting area, one to keep the eyes on. I certainly have seen an increase in the number of early stage companies addressing this and bringing this technology. And it’s completely orthogonal to optical imaging. It’s acoustic. If you think back to high school, you know, transverse and longitudinal waves, the two are completely separate, but will also require its own processing and understanding of the data that’s coming back.
FWM: That’s right. And the picture has to be created; it doesn’t live in the real world. So it has to be created from data, which is one more step. But the way processing has advanced, it’s probably not much of a factor holding back anything.
Yates: I think the processing would be less of an issue. And I’ve seen acoustic systems. When you think about visible imaging with a typical camera, we’ve got an extremely well-developed ecosystem producing the image sensors. Think of Sony or OmniVision, for example. The amount of time and effort invested to make excellent image sensors, that level of resource and applied expertise has not been pushed towards developing acoustic imaging sensors in the same way. And the actual sensing devices do not exist in such a detailed ecosystem.
FWM: Actually, that, that brings us to another question. Let’s agree that machine vision is a well-developed and, comparatively, quite an old industry. I feel like it’s growing into other areas like security, fire prevention, those types of things. Can you steer us and show us where else are we finding those types of solutions?
Yates: Yes. To put it into context, we often think machine vision and then think factories. Now, we are starting to use the term “vision tech” to capture the fact that vision technology is used well outside the factories. You’re exactly right: vision itself as a technology and as a market is essentially a horizontal market. So it serves many different verticals. Obviously, it’s very strong in factory automation, and for inspection of goods, I mean, those are natural places. But also in defense, and if you think of all of the observation systems around it in agriculture, increasingly, whether sitting on a drone and mapping out exactly how your field is working in terms of your plant growth, also very important in medical.
If you ever go to a hospital, you’ll almost certainly have many tests. There will be some type of vision element within your visit, either scanning you or in the lab scanning the samples that have been taken. It’s almost ubiquitous, it’s very rare that you would have a medical facility without a vision-based system helping out.
This is part of the reason that there’s such strong strategic interest in machine vision, or rather the vision tech industry—because it feeds so many verticals. And ultimately the value proposition is, I can do something that’s very similar to what a human does, which is seeing. I can do it automatically, I can do it fast, I can do it cheap, and I can process the data coming out to get useful information. And that’s a pretty strong value proposition.
FWM: This is the second time in a month where I’ve run across a conversation that says the old advice of fast, cheap and good, pick two, doesn’t hold anymore. It just doesn’t hold. You can have all three.
Yates: That is a good way of phrasing it, Dave. Exactly that. Vision provides one of the highest density data forms that exists. Just on your phone, you probably have 24 megapixels of data that can be captured at 120 frames per second. So you capture this and see, of course, that it then it has to be interpreted to get the information you want out, whether this is a photo and it happens to be my daughter and her friends, and that’s what you want from it. But the sensing capability has got a tremendous amount of information stored within it.
FWM: Agreed! The advancements are fast and as you said, ubiquitous. You see areas of growth in the startup world bubble up from the inventors, small research firms, or incubators. What trends are you seeing from these technology birthplaces?
Yates: My background is in startup companies; that’s more or less what I have done since I left university. So I have to say, when I go to a trade show, I always go to the startup areas. There are always startup areas at these events. When I was working at Rockwell Automation, you’d see this would be the place to go because this is a place where you can see little hints of innovation, or it’s easier to see the innovation that’s happening.
At one level in the industry, we have consolidation happening. There are a lot of transactions and some islands of consolidation across technologies and portfolios. This is balanced by the number of companies that are just coming into the market, seeing it as an attractive market, raising finance from investors in order to go into it. They are selling their proposition and bringing new innovations and technology to the market. So I see it as, as fantastic and a very positive sign that there’s a strong, vibrant industry there and not all the problems are solved, and we know this. There is a lot of opportunity out there, it’s fantastic. And to be honest, it’s great fun to be in the startup areas.
FWM: It is a great place to be. If I may ask one last question: So many of my followers are in fabricating or fabricating and machining. They read my content, they hear me talk about this stuff. They hear you talk about it, and it’s very attractive. What is one piece of solid advice that you can give them as they’re looking at potential solutions to their manufacturing problems?
Yates: That’s a very good question, David. Probably the one thing to do if they’re starting from the beginning on this is vision technology is to realize it’s still more complex than it could be. I would certainly advise speaking to a competent expert in the field, someone appropriate for the problem they’re looking at, whether that’s a system integrator, a distributor or channel partner that’s got significant experience, or somebody like in Vision Ventures’ position to be able to, to really talk through what it is that they’re looking to achieve and have an open and practical conversation about the technologies that are there and how it all should fit together. And one could expect a very good response and a very honest and open approach if the company or expert are approached for the first time by someone saying, for instance, “I want to use machine vision, but I don’t know where to start.”
FWM: That is very good and very practical advice; there is a lot of expertise available out there if you look for it. Thank you, Chris, thank you so much for your time today.
Yates: No problem. It’s been a pleasure to have chat.