Industry of Things is moving into production and manufacturing and will change the industry in the long term. Everything is networked in the factory of the future. Machines, robots and people communicate digitally. The fourth industrial revolution is still in its infancy. It needs many steps to implement the vision. Artificial intelligence is the precondition.
Prof. Dr.-Ing. Torsten Kröger, Head of Institute IPR, Karlsruhe Institute of Technology:
"The benefits we see today are those that come through the sub-field of artificial intelligence - machine learning. And machine learning means that we have software algorithms that learn from data. And the most important thing at this point is actually the Data. Unfortunately, in many cases we do not have enough. It is also very difficult to gather them and for this it is simply important to create standards in order to obtain this data at all and to allow new applications."
Automatica 2018 shows what new applications look like when real production worlds provide virtual data for further processing. Fanuc, for example, connects 68 machines and robots to each other on their stand. Production can therefore be tracked in real time, providing transparency and predictive maintenance before machines fail.
The company teamtechnik is also focusing on making the individual components of a factory smarter. These data are collected with a special software, standardized, securely and efficiently.
Farid Nasimzada, Product Manager R&D Software, teamtechnik:
"Ultimately, data collection and analysis of data is used to improve my product quality, identify weak spots, and then to correct and improve. And today I just need a tool that supports me. Not to have to do everything manually, but a tool that will take my work off of me. It takes a lot of time to carry out such analyses manually. We can supply, prepare and make available all the information in advance.”
In order to avoid error sources, Siemens also relies on digitization, more precisely on the digital twin of a real machine. With it, changes in the production process can be simulated and tested in advance.
Tobias Fengel, Marketing Manager Siemens:
"You want to import a software update into your machine. You want to make optimizations on your machine, make changes, insert new components. That means I have to carry out an engineering and so I need to shut down the machine, I have to import my software, I have to test it and the whole time the machine is at a standstill. It cannot produce. But the end customer's goal is basically, the machine has to be running, it has to produce. And how can I shorten these downtimes? I take the digital twin, I test my optimization work in advance at my office desk. And only when I am sure that what I want to import into the machine, that it actually works, only then do I go to the customer and these downtimes are significantly reduced. "
To make a production process flexible and easier, manufacturers are increasingly relying on artificial intelligence. One of the aims of this is to make robots more easily programmable in the future, and to have them learn independently via image recognition, for example for sorting processes.
Michael Bellmann, Application Engineer MVTec Software:
"One must provide it once with a large data set of images, telling it this is an object A, B, or C. The neural network, which works a bit similar to the human brain, can now recognize the objects independently. And the objects can also occur in different layers."
The neural network of the robot learns by itself. This section of machine learning is called "deep learning".
Automatica 2018 is the platform to discover the world of tomorrow in robotics and automation to smart components and digital data analysis.
This media asset is free for editorial broadcast, print, online and radio use. It is restricted for use for other purposes.