Audi Production in Germany – Audi Smart Production

The assembly line has set the pace in automotive production for over a century. Now it is increasingly reaching its limits. Numerous derivatives and individualization options are making products more and more varied. In a rigid, sequential process, that complexity is more and more difficult to master. In the Audi Production Lab, project manager Wolfgang Kern’s team is preparing modular assembly for mass production.

Audi is initially implementing the concept in interior door panel pre-assembly in the Ingolstadt plant. In the test operation, work no longer follows a uniform sequence. Instead, it meets particular needs. Automated guided vehicles (AGVs) bring door panels right to the station where the components are assembled. In addition, we can link the modular assembly to specific production steps. For example, now, a single worker can install complete sun blinds. That used to require two or three workers due to the preset processing times in an assembly line. Another major advantage of the flexible system: Audi can employ people who can no longer work on the line due to physical limitations.

A model’s various designs and equipment variants can be examined quickly and efficiently in different environments and lighting conditions using virtual representations. One urgent goal is to move the design into production with the least number of possible cuts and get it on the road for customers. To that end, experts look at design drafts from a model’s early development phase up to the tools’ release to see if they can reproduce them in serial production. Final approval of the cars’ surfaces comes in what is known as data control milestones.

The most important tools for doing that are big screens, known as powerwalls, which allow us to depict a car in its original size. In combination with the visualization cluster – a computer cluster with a total of 26,000 CPUs – cars can be presented realistically and with physically based light, shadow, and reflection calculations. That process is the basis for the virtual decision-making process.

This technology is also used in tolerance management. That way, Audi ensures it can build a particular model to spec from both a constructive and a qualitative perspective. With 3D simulations of the body, the effects of component and assembly tolerances can be foreseen in the vehicle's image. The simulation results are then realistically visualized using virtual reality. That way, the experts from Audi Production can influence the design and development process regardless of time and place at almost no additional cost and, for the first time, from the production facilities.

Virtual assembly planning
Virtual assembly planning not only saves material resources but also makes innovative, flexible collaboration possible across different locations. It eliminates the need to build prototypes in the planning process. A scanning process generates three-dimensional point clouds that can be used to virtually reverse engineer machines and infrastructure. The software is based on artificial intelligence and machine learning. It makes it possible for employees at Audi to navigate through assembly lines virtually. Volkswagen’s Industrial Cloud gives them an efficient tool that allows them, for example, to compare locations and use appropriate solutions from other production lines in their planning.

Right now, Audi is working with NavVis to test Spot the robot dog so they can do the 3D scans as efficiently as possible. Around four million square meters (43 million sq. ft.) and 13 plants have been involved since site digitalization started in 2017. Scanning 100,000 square meters (1,076,391 sq. ft.) – for instance, in Audi A6 production in Neckarsulm – takes about three weeks in single-shift operation. The scans can only be done at night or on weekends.

By contrast, Spot the robot dog can do that scanning in 48 hours and figure out his route autonomously.

AI in production
Artificial intelligence and machine learning are core technologies in Audi’s digital transformation and modern production. An AI algorithm in the Ingolstadt press shop helps identify flaws in components. That procedure is supported by software based on an artificial neural network. The software itself identifies the smallest flaws and reliably marks them. The solution is based on deep learning, a special kind of machine learning that can work with unstructured and high-dimensional data volumes.

In another pilot project, Audi uses artificial intelligence to check the quality of spot welds in high-volume production at its Neckarsulm site.

___

Video Timeline
0:00 – AI Crack Detection (Body Shop)
2:18 – AI Underbody Cover Inspection (Assembly)
3:45 – Quality Detection in the Resistance Spot Welding Process with AI
5:14 – Modular Production
8:11 – Virtual Geometry Validation Process (VR)
11:17 – Virtual Assembly Planning (VR)
12:39 – 3D Factory Scan (Boston Dynamics 'Spot' Robot)

[dcasteaux]

Having owned several VAG vehicles over the past 30 years what's really impressive is Audi's *Built In Obsolescence Team* who'll ensure the car is scrap after 8 years. From plastic gears that get brittle to mild steel subframes coated with the thinnest paint layer so they rust through ASAP it's all done for one reason: to get you to replace the car sooner with another new one.

[turilarosa]

I really like your videos, but why you do not also register somebody that describes the processes? (also a vocal background like "how it's made" tv show)