At 7th TPV Innovation Day, we had the pleasure of interviewing Dr. Myriam Jahn, CEO at Q-loud and one of the leading IoT experts. Read on to learn more about her view on the industry 4.0.
You are one of the key experts in the industry 4.0 with a lot of experience in IoT. How did you become interested in this field? Has technology always been your thing?
Yes, it has. My father had a sawmill, a very small one with 10 employees. As a small child I was playing around the machines that my father had constructed all by himself. The industry 4.0 was my field before the term was even invented. What happened? First, I studied information technology and after getting a scholarship I switched to business administration. I’ve always been a little bit sad to have given up information technology studies. My PhD was something between business administration and information technology. I worked for an automation technology company for 15 years and during that period I obtained the title Master of Science in electrical engineering. So, my expertise is information technology, business administration and electrical engineering.
How did you start with IoT?
I always wondered, if there are so many sensors in the machinery, why don’t we obtain data from machines in order to help with production planning, manufacturing execution systems and consequently help us understand what happens on the shopfloor? In 2012, the term industry 4.0 came up and I thought that was the exact answer to my question. That was the starting point for me. It started with a vibration sensor that wasn’t necessary for the machine as it was only a condition monitoring sensor. The trouble was we couldn’t sell the sensor. There were two reasons: first, it wasn’t mandatory for the machine, and second, the maintenance people at the production site said they didn’t want lights blinking green and red, they just wanted data on the computer monitor. That was a nice wish, but it cost a lot of money. At the time, the connection to the monitor cost 25,000 € and the sensor was only 250 €, so it was a big issue to integrate it in to ERP or MES. The same issue was with 6.5 € inductive sensor where the connection and integration also cost from 25,000 € up. We noticed that MES was the meeting point and decided to buy MES software support to gain the knowledge to sell this vibration sensor. Then industry 4.0 came and everyone was in awe how we had managed to do that. It was then we started to sell software. We had a successful start, because we sold parametrisation software. We had previously sold hardware with free software, but then we had to start charging extra for the software. The margins were high and we were successful with the sale. The company has been doing very well and has now over 300 employees, but for me it wasn’t a challenge anymore. The biggest challenge I saw was to get the companies to implement cloud. So, I started with QSC, the mother company of Q-loud, that was in a line of business interesting for cloud integration. That was the starting point of Q-loud.
What would you say are the benefits of IoT for manufacturing companies?
There is a discussion going on whether digitalization replaces a lot of people. I don’t believe that. I believe that collaboration with the robots gets easier and that people on the shopfloor can make the decisions by themselves with digitalisation. This is actually more of a threat to the management than it is to the shopfloor. There is proof that productivity of the management will go up because if you get the collaboration right, there is no need for such a number of managers. In the end, the biggest benefit will be an increase in productivity.
Should the implementation of IoT be one of the first priorities for the company and if so, why?
For smart factories and industry 4.0, I think it’s crucial to get all the software on cloud. A cloud-based software should be one of the priorities, followed by application of artificial intelligence (A. I.). A.I. is all about learning, and learning means you need some time to learn. Time is crucial so the companies that implement it first have an advantage. For a company like TPV, where core competence is manufacturing and development, the key priority is also crucial to get all the data out of the machines, which isn’t easy as they have different output data. You need to bypass the PLC and have storage on the machine. You should have the possibility to change algorithms and get the most of the data.
Does IoT present new risks and possible issues?
Starting with the cloud, the data ownership is still not really defined as there is no such thing as a legal data ownership. This means that if you store information in a data center, the data center is officially allowed to take your data and use it. The data centers are secure, but legally they own the data because by current law: who stores it owns it. For this reason, you need to address this with contracts. Then we come to the issue of A.I. If you use an open source A.I. and don’t take care of that with your contract, then they also own the A.I. If you have a product like AGV (automated guided vehicle) and you have digitized connection to the cloud, then it is also crucial to have a certain portability with the cloud, so you can migrate at any moment if necessary and they can’t lock you in.
How is the data processed and how much maintenance is needed?
First of all, you shouldn’t store all the data in the cloud, because it’s too expensive and there’s too much data. What you should do is pre-process the data at the site or preferably on the machine. It’s crucial that whatever is on the machine should be in total access of the cloud. This way you get only the crucial data into the cloud. I can give you an example why you should have processing on the machine. You might be familiar with the fatal accident of the uber autonomous car. The uber car driver didn’t pay attention, but the car detected a woman with a bicycle and a bag. At that point, the car couldn’t decide whether it was a human or a bag. The core A.I. was stored in the cloud and the car had only algorithms. The car couldn’t make a decision and asked the cloud what to do. It took 6 seconds to get the data across from the car to the cloud, which was too long. The point is A.I. should be in the car and you should be able to upgrade the A.I. in the car from the cloud. The decision-making part of the A.I. should also be in the car.
What is your experience with companies that have implemented IoT? Can you provide examples of good practice?
The cloud approach is used by Techem, a big supplier of heating monitoring. In apartments for rent, you need to monitor who is using how much energy for heating. They have about 2 million devices in the apartments, all in the cloud, because you have to have all the data distributed. The cloud is done by us (Q-loud). They will also implement more intelligence on their devices and we will be a part of that. We have a lot of construction companies with precision tools which have installed beacons in tools in order to receive the information whether it is on or off and where it’s located.
In what way could a company like TPV make use of IoT?
Firstly, I’m very impressed with TPV. I got the impression that there is a can-do mentality which companies often lack. The leap that you have already made from manufacturing mechanical parts to AGV is huge and incredible. You are a perfect example of all the benefits. There are a lot of synergies between R&D, manufacturing, and developing digitized products, which is essential for a smart factory. You could benefit by digitising the AGV a little bit more. Connecting the machines to the cloud will be the next challenge.
How can we change the mentality of people and prepare them for work in smart factories?
I think that is the easy part if you give people transparency on what their machine is doing. Normally, the workers are a team with their machine and you need to give them transparency what happens with their machine. That way, they can decide what steps to take next.
I can give you an example from the shopfloor. I made a presentation for machine tool builders and told them it shouldn’t take more than 6 months to learn how to handle their machines. They laughed at the time. They aren’t laughing anymore, because the machines should be easy to handle and the user interface should be fun and easy to learn. This is a key factor for people to accept it. Working with the machine should be fun just like applications on our phones.