The autonomous transport robots are camera-based monitored and can be positioned with an accuracy of less than half a millimetre. At Volkswagen's Smart Production Lab in Wolfsburg, the software and algorithms behind the control system are developed.
The autonomous transport robots are camera-based monitored and can be positioned with an accuracy of less than half a millimetre. At Volkswagen's Smart Production Lab in Wolfsburg, the software and algorithms behind the control system are developed.
( Source: Roland Hermstein)

Smart Production Volkswagen Smart Production Lab: Advance development for production

Editor: Florian Richert

Engineering, Pilots, Prototypes: At the Smart Production Lab in Wolfsburg, Volkswagen develops production and logistics technology that may be available in production in a few years.

Wolfsburg, Smart Production Lab. The team consists of 38 employees, all engineers, developers, or software specialists. Their mission is to devise technology for production and logistics that Volkswagen can use in its factories. The challenges are high: "For about four to five years, the development around the term 'digitization' has accelerated considerably," describes Martin Hofmann, Head of IT in the Volkswagen Group. "We have 120 plants worldwide, plus our dealers. They all work with digital information that has to be controlled. Besides, there are around 14,000 IT employees - all of whom need IT support. At the same time, the environment is changing massively: new business models, car-sharing, and e-mobility are just three examples".
Three years ago, the automobile manufacturer, therefore, began to set up its so-called Smart Production Labs. Each location deals with different topics. What they all have in common is that employees can test and try things out here, even at the risk of a project not working. In the labs, we can span the arc," describes Hofmann, "without endangering ongoing production. But the developers work closely with colleagues from production, "who tell us whether something makes sense," explains Matthias Behrens, head of the lab in Wolfsburg.

Gallery with 5 images

In addition to the Wolfsburg site, there are other labs in Barcelona, Berlin, Dresden, Munich, and San Francisco, each working on different topics. The teams in the labs are international: in Berlin, for example, the employees come from 52 nations. It is more vital for us to develop products in the existing labs that are useful in production," says IT boss Hofmann.

AI for robots and transport systems

In Wolfsburg, Behrens and his team and partners are primarily developing artificial intelligence for robot controls and autonomous transport systems - close to real production. However, the developers have now gone beyond the experimental stage: they develop concrete products for the factories, application-oriented. Behrens asks, "What helps the employee on the line?" describing the focus of the development. At the "Open Lab Day" in September, the team showed some of its progress.

One application is already well advanced: object recognition, which checks a set of components for completeness at goods receipt. The app means that the software uses cameras to monitor the incoming goods basket and recognizes what kind of parts are in the container, such as a set of safety belts for a five-seater car. If a part is missing, the system reports this - an employee can intervene accordingly. The system has been testing Volkswagen at a picking station since the beginning of the year, with astonishing results: the number of missing parts on the line has fallen by 97 percent.

The need to catch up is more in the mindset area.

Ralf Brunken

Predictive maintenance - self-developed

Another project deals with predictive maintenance, or predicting when a machine will fail. "We talked to the staff at the plant to find the right data and draw the right conclusions," describes Matthias Behrens. The gluing system, as an example, the absolute value was the maximum adhesive pressure. The lab staff themselves developed the software behind the operation. "It's a mixture of artificial intelligence and statistics," explains Behrens. "The advance calculation is a statistical value; the process is determined by artificial intelligence.
The algorithm requires only a small amount of failure data to learn," Behrens continues, "which sounds simpler in his explanations than the development in the doctoral thesis of a Mexican team member might have been. A further advantage of the development is that the algorithm is not tailored to this one system, it can be adapted to the particular requirement."

Protecting personal data

Another in-house development is the so-called "identity shield," an automatic pixelation for video recordings within production. According to the applicable necessary data protection regulation, no personal data may be stored unless there is a particular reason for doing so.
The software recognizes persons in the recorded material - for example, from process monitoring or picking - and creates a kind of shadow over their outline. The video material leaves the camera anonymously. The basis for this is, of course, mature cams, "and to test the various models, we have also developed our tool," the project team described with evident pride.

The Auto-Uni

All these developments have a basis: skilled personnel. "We need the right people," says Ralf Brunken, head of Autouni, where Volkswagen is building up its own IT specialist staff. To be accepted into the "Faculty 73" program, academic training is not necessarily required. "We are more interested in whether the applicants are cognitively capable of IT and whether they have already dealt with IT," says Brunken.
In March of next year, another 100 students are to begin their further training at Autouni, and another 100 are to follow in September; another 200 from 2021. Above all, we need software developers," says Brunken. "Around 4,000 to 5,000 in the next five years" - to increase the depth of added value in software development. The higher education landscape in Germany cannot cover this because universities already have a capacity problem or other priorities.
"Demands in this area are growing much faster," says Brunken. "The `spirit word' artificial intelligence alone - which, in my opinion, is hype and should be demystified - makes this clear. For Ralf Brunken, there are also social reasons why other countries are still here: "We must become more willing to take risks. So far, we have been very strongly driven by engineering. Everything must be deterministic and plannable. It would be better to take a risk at an early stage. Instead, we make a plan - and we keep it that way. We don't have the pent-up demand at the scientific or expert level, but rather at the cultural mindset," explains Brunken. That's why Volkswagen - like many other companies - continues to involve foreign specialists who come primarily from India in the software sector.

New teaching methods

The "Uni" in "Autouni" dates back to its foundation in 2004 when the university was to be accredited. Nothing came of it, but "we don't need our university either," says Brunken. "We have a network of more than 500 people who we can acquire for new topics.
All in all, however, education itself is undergoing a `digitalization change,' Brunken says. "Not only is the content changing, but also how we train has to change," says the head of the educational institution. "We use much more strongly online formats, and we interlace more so that humans can help themselves mutually, describe Brunken, and clarify it with the example Google: Google makes 80 percent of its training courses Google to Googler. That means: Experienced colleagues train new colleagues - and have the time for it. According to Brunken, free space instead of advisers - exaggeratedly represented - is likewise a cultural change and one of the most exciting tasks, which I see at present in the enterprise: all on the part of the further training to accompany.

Industrializing developments

Another exciting task will be to transfer the developed technology into real production. At present, the developments are still limited to the laboratory environment except for tests. But: "We are far beyond the proof-of-concept character - and on the way to bringing our concepts into an application and industrializing them," says Matthias Behrens.
But there is one hurdle: "Of course we can, for example, develop intelligent robots very far. However, the use of the `24/7' line places even higher demands on the technology. Besides, it must be clear how the cooperative deals with it before it is used in practice. Is the topic understandable for labor law, can it be `cast' in permits? Keywords: operational safety, occupational health, and safety - here the technology are sometimes further than the approvals."
It remains to be hoped that the cultural change mentioned will begin soon so that Matthias Behrens and his team can accompany their development work in production.

This article was first published in German by Automobil Industrie.