COMMON VISION BLOX

Mr Keppler, the merging of the worlds of control and machine vision is a current trend in automation. What is the latest situation here and do you forsee any difficulties?

Peter Keppler: Amongst other things, the fact that leading controller manufacturers such as Beckhoff, B&R and others are pursuing the subject of machine vision with greater dedication than before shows that these two worlds really are moving closer and closer together. There is now a general conviction that "seeing" plants can do more than "blind" plants in many branches of industry. As supplier of machine vision technology we naturally view this development as a logical and very positive step. Machine vision is now an established technology in the field of automation and with regard to Industry 4.0 it's completely indispensable.

As is so often the case, however, the difficulties along the road to user-friendly solutions lie in the detail. Important questions here are, for example, cross-manufacturer compatibility and communication between the controller and the machine vision system. There were no comprehensive standards for this topic in the past and that has unfortunately continually delayed the merging of the two worlds. Good standards aren't simply the "lowest common denominator", but must continue to allow differentiation between providers.

In the meantime, there is growing awareness on both sides that the future lies in standardisation and compatibility. With the latest developments and the publication of the OPC UA Companion Specification for Machine Vision, we are currently at the dawn of a new era: this standard will considerably accelerate the interaction of machine vision and automation across all manufacturers and in this way make a big contribution towards making "seeing" plants an actuality according to the concept of Industry 4.0.

Manufacturers of classic sensor components are increasingly offering imaging products. Why is that?

Peter Keppler: The reason is clear: users are requesting them! Many current Industry 4.0 requirements can already be solved in an elegant way with quite simple machine vision solutions, whereas the classic sensors do not offer the necessary flexibility. The selection and operation of these "vision sensors" is in some cases so simple that users can also be served efficiently through a sensor distribution network. However, I urgently recommend users to make use of independent consulting and training courses in order to ensure that they actually find the ideal solution for the individual task.

What effects is the merging of the two worlds having on their users?

Peter Keppler: In my view, automation companies must concern themselves more with the subject of machine vision, as they will no longer be able to meet the requirements for Industry 4.0 in future without this technology. Machine vision systems must be simple to operate for this new group of users and must take into account their level of knowledge. However, this requirement is at the same time a big challenge, because the bandwidth of machine vision applications is very broad. On the one hand it is advantageous when more and more applications can easily be solved by machine vision devices that are simple to operate. On the other, focusing too strongly on "simple applications" also hampers the development of really innovative solutions with a unique character.

What ways do you see of driving the use of machine vision into new task areas?

Peter Keppler: An important catchphrase for this at present is ‘machine learning methods’. We have concerned ourselves with this subject for many years now and have already been able to successfully complete thousands of projects with technologies based on it. Thanks to the enthusiasm that has developed in recent times for the term Deep Learning, the use of this technology for machine vision has now become popular on a broad basis, even though Deep Learning – that means multi-layer neural networks – is not ideal for industrial imaging in my opinion. With the right machine learning algorithms, however, we will soon see high-performance, flexible vertical solutions that will even run on inexpensive embedded systems, making extremely cost-effective systems possible.

Another way of simplifying machine vision, of course, is to use graphical user interfaces, from which programmers and integrators above all could benefit through fast induction and short time-to-market phases. However, I urgently recommend attaching importance here to vendor-neutral hardware support. In the field of camera technology, the established standards GigE Vision and USB3 Vision already offer a good basis for equipping oneself for most applications. Additionally OPC UA will establish itself for independent communication with the plant controller.

Let's go back to the merging of automation and machine vision: how do you asses this development in summary?

Peter Keppler: There has been a consensus of opinion for years that the combination of these two worlds offers considerable advantages for automation users and can create innovative possibilities. In the era of globalisation and Industry 4.0, the close connection of these key technologies is an important prerequisite for cost-effective automation solutions. With the current developments around the OPC UA standard, a large step has now been taken in my opinion towards the significant simplification of integrated solutions consisting of machine vision and automation in the interests of the users. We make an important contribution to the simple, risk-free entry into machine vision with our independent technology training courses for the planning and design of machine vision solutions.