AZoOptics speaks to Allied Vision at Vision 2018 about the differences between machine vision and embedded vision.
Who are Allied Vision and how do you fit into machine vision systems?
Allied Vision is a global player, we have locations all over the globe, three locations in Germany, and one in Canada, two sales and support offices in the U.S., one in Singapore and one in Shanghai. We're able to support our customers around the world, for a full working week, 24 hours a day and combining all those factors I think we're an important player in the market and well-known in the machine vision field for, the reliability of our cameras, quality and being at the forefront of innovation.
You're showcasing a new product here at Vision 2018 what are the basic principles of it and how did it come to production?
We're introducing the Alvium Camera Series at the show, our new, innovative camera platform, which is aimed at the limitations we see with current embedded and PC based vision systems. It offers the designers an unprecedented level of freedom when designing their system, it is a camera platform that tries to bridge the gap between embedded vision and machine vision. As we are from a machine vision background we bring our machine vision experience and know-how into embedded vision because that's a big trend that we're seeing.
What's the difference between an embedded vision system and a machine vision system?
Currently, in the machine vision world you have a camera and a lighting system, and then you have a long cable, and it connects somewhere to a workstation or a PC station, where the analysis is done. The machine and the camera system takes the images, and then there is a PC based station away from it which does the image analysis. At those stations there are just regular PC systems as but they're quite large, they're power hungry, and you have cables running all through your installations and your factory floor which brings some limitations that embedded vision will alleviate in the future.
What are the limitations that you believe embedded vision will alleviate?
The big PC stations which consume a lot of power will become obsolete, embedded systems are much smaller, and consume much less power which also gives customers the opportunity to put the system directly onto the machine because they're so small. This smaller size means that there is no need to have the cabling all through your factory floor so the system takes the images and processes the images right there, in the same spot and I think that is the future and how it's going to be.
What led you to bring out the Alvium Series, was it more of a hardware issue or a software development?
It's a combination of both but it started with the hardware for Allied Vision as that’s where our expertise lies. The hardware needs to be able to do certain things, which means that there are certain pre-conditions that need to be met with it. First of all, the hardware needs to be small and it needs to be energy efficient so that customers can really put it everywhere, we've really tried to address this by developing our own vision processor, our LVM technology. The LVM technology is an image signal processor with an extensive image library which enables customers to really do a lot of processing on the camera and get a nice crisp image directly from the camera.
Is the image processing library integrated in the ALVIUM technology similar to a neural network where the customer will be able to program it themselves?
No, that will still be done externally to the camera on an embedded board, for example, one big player is Nvidia. They have dedicated chips especially suited for processes such as neural networks or deep learning. Our camera can do pre-processing on the images so you can get a good, sharp, clear image from the camera that you can process later out of the camera. Those calculations to help you get that image, however, can be done on the camera, and don't need to be done on the embedded systems where all the processing power is available for things such as deep learning or neural networks.
What industries do you think will benefit the most from this technology?
We come from machine vision and machine vision is used in a lot of different industries already as seen through our customer base. We can already see the benefits that it will bring to the machine vision industry, but, it goes far beyond it. Now the Internet of things is a ‘trending topic’ creating vision systems that are smaller and smaller will open new opportunities within that area. This technology won't be limited to specific industries it will spread into almost any industry where there is an inspection task, or you need to look at something. Cameras are everywhere so it's hard to pin it down to certain industries.
What application has Allied Vision used this technology for already?
As this is the launch we are only just bringing in some of the applications, but we have certain demos which we have at the show. A very basic example is in identification tasks where you need to identify a certain object or a certain shape. Within that example, there are certain mechanical parts being produced, a milling machine which produces the parts and then a vision system on a robot which inspects the parts from all sides. With our new system being so small, the embedded system or the processing can be done on the robot where the camera sits.
Definitely, there needs to be communication between the embedded system or the operating system running on the embedded system and the camera, it needs to be able to control the camera, when is an image recorded it needs to know under what circumstances was it recorded, and what were the settings. Furthermore, the image needs to go to the host in a controlled manner and there might be some compression involved in this. The two things must speak the same language, not only is the hardware important in the new systems but the software also plays a crucial part. It becomes more important as more and more specialized processors are used on embedded boards the integrated systems used on those embedded boards are very dedicated and they have different processors for different tasks. This means that the camera also needs to speak that language to be able to benefit from those specialized tasks that are required. Allied Vision believes that software will become more and more important as we progress but it will always be a mix of hardware and software here.
For more information please visit Allied Vision.