‘Data does not automatically become information’
An interview with Dr Gunther Kegel, Chief Executive Officer, Pepperl + Fuchs GmbH, Germany.
Pepperl + Fuchs GmbH
Dtata gathering and stripping
Dr Kegel, you made an interesting observation about the data deluge. Can you explain?
A huge amount of data is generated today in a typical process and not all of it is used to generate information. It depends how we connect the sensors that are the sources collecting all the data, how do we pick up the domain specific data from that and make it more or less structured general data that can be used in a harmonised way. Tools like Google analytics and the deep learning algorithms work on such stripped off data – which then is no longer domain specific but becomes general data – to read and analyse the values of various parameters so one has to ensure that the correct data is available that meets the expectations of the big data analytics software.
Does this mean that Google is likely to become a major player in automation?
Some consultancies have claimed Google will be the competitor of automation companies in future, but that is unlikely. Google does not want to build sensors or transmitters or safety devices, but what Google would like to understand is the nature of domain specific data that we are handling, how to turn that into general data that can be useful for big data analytics and maybe develop new human machine interfaces, etc. This is where we expect Google will become partially a supplier for companies like Pepperl + Fuchs. On the other hand, in generic data they can even become a competitor, as in that area they can do much better than us, with a huge number of people working on that. A collaboration model is more likely to emerge, because we are the only ones with the capability of taking the domain specific data and turning it into generic data, and that is what Google and the SAPs of the world lack.
Do you see a Google like system emerging for the automation industry with this collaboration model?
No, I do not see that happening and the reason is simple. The automation companies do not have policies that justify such investments. We need to take technologies from other sectors such as communications, IT, mobility, which have economies of scale and turn them into cheap solutions by taking the available technology from them. If we start doing a Google for automation by ourselves, the quality, amount of effort, resources and investment will produce something that is just half way of a solution, so I strictly feel this is not going to happen.
How do you make a distinction between data and information?
A very interesting question! Data is nothing but a set of computer relevant information, for example, regarding one parameter or measurement point, in terms of reading and value; it tells you about where, what and when, but nothing beyond that. For example, it will tell you a certain degree of temperature at a certain point of time – essentially what we call data. Information, on the other hand, is that this certain temperature at that point of time should have been 10 degrees lower, which means something is wrong. Only when we put several such data together can we get information. But there is no sharp line dividing the two as for some people and processes, the data itself is a lot of information as they have enough background knowledge, but for others, only combining several such data becomes information. Data does not automatically become information – just data combined together is not information. Data is what is generated by the devices; information is the first level of computational application, the deep learning of big data.
With so many vendors, devices, data, information and applications, how does a typical industry user make the decision about the appropriate decision making tool?
It is about value. Does a new application or software generate value for me? Can this data turn into information that I can turn into value? Does it generate profit for the investment made? The nice thing about these applications is that they are relatively inexpensive and without the operational hassles, sitting on a cloud of the vendor, and there is a set of tools one can use over there through the smartphone or some other device. It may not work 100% but the 90% it delivers generates value. The IT guys have completely new tools, nobody is trying to develop monolithic software monsters like their own version of SAP or stuff like that, but just the service architecture underneath that provides the necessary operation systems are doing a good job in delivering simple, need based solutions, very lean apps that address a single feature and there could be hundreds of such applications sitting on the smartphone. Is this a problem? Not really, because every application has a single purpose, it uses a small interface, sits on a cloud and you are much more agile and it costs little money, it is really very cheap. Because you do not buy a cloud, you just buy a service that is accessible from anywhere in the world.
So basically it is all about accepting the change to move ahead amidst the disruption that is happening?
Most companies that are currently engaged in automation products business carry a legacy, have an installed base of a successful business and are running profitably. For such companies changing or interrupting or disrupting their existing business model is simply not possible, that would be like cutting their own veins and bleeding, which does not make any sense, so they are forced to continue, but at the same time many of these companies are doing the new stuff on the side, by launching new start-ups or investing in joint ventures. While leveraging their traditional strength in the existing business, they invest into new technologies, which over a period of time will leverage that expertise – that is how most conglomerates do it as they are aware icons of today may disappear tomorrow. I have never seen any company successfully transform overnight from an existing model into a new business. So the transformation will take time as the new model is built along the existing one, which will then fade gradually ensuring a smooth transition.
About Industry 4.0 creating a substantial impact, data from VDE – Germany shows a wide gap in the expectations from companies (6%) and universities (25%) till 2020. Why so much divergence?
Well, the universities are more optimistic because for them, if a model works, it is already done, whereas for companies, they have to consider the practical aspects, it is just the beginning that the model works. These nice technologies are available but not fully operational, do not have the necessary reliability or the costing, they do not have the necessary business models and the support structure so it takes time. While universities are always 2-3 years ahead, industries need that much of time to realise what comes out of universities, fine-tune the entire practical and compliance issues and only then make the transition.
If the German companies expect only 6% impact of Industry 4.0 by 2020, what could be the case for India?
Presently difficult to say in percentage, probably not much, but the advantage India has is on the software side and there is a tremendous opportunity here in terms of deep learning and big data. Also for the first time, here is a technology that does not call for huge investments on the ground as unlike manufacturing, software development does not need that kind of infrastructure and India has the proven ability here.