Audi “FelGAN” uses AI to inspire new rim design

Using artificial intelligence (AI) in all divisions is the goal Audi has set for itself on the road to becoming a data-driven company. The arrival of FelGAN means that Audi has successfully deployed artificial intelligence software that opens up new sources of inspiration for designers.

Creative people are always on the lookout for inspiration. The same goes for the designers of the new wheels at Audi’s design studio in Ingolstadt. But where to find untapped sources of inspiration?

Designers take inspiration from FelGAN

To “jump out of the box” is easier said than done. Because in the creative process, people always tend to resort to what they are familiar with. It is precisely for this reason that the Audi IT department and the Audi design team have developed the artificial intelligence-based FelGAN software within the company. Today, the program enables innovative designers to draw from an almost endless library of ideas. With this software, designers can explore design drawings from a new perspective and get suggestions that can be further diverged and put into practice. Specifically, FelGAN can quickly and automatically give a large number of realistic design pictures, and can also recombine existing designs in a targeted manner.

FelGAN is jointly developed by the Audi IT department and the Audi design team

FelGAN has thus become the creative center of the Audi rim design team. Teams can communicate and share new ideas and versions with each other on the software, and can conveniently explore various design possibilities of parameters such as shape, color, and surface structure in real time.

How did artificial intelligence learn to design the wheel?

“FelGAN” is named after the German word “wheel hub” (Felge) and the acronym “GAN” for Generative Adversarial Network. Generative Adversarial Network (GAN) is a special self-learning computer program, which contains two algorithms that play games with each other, alternate iterations during training, and continuously optimize.

Designers use FelGAN to polish individual elements so that they blend harmoniously with the overall design

The specific working principle is: one of the algorithms acts as a “generator”, which is responsible for generating images of a specific subject, which is the hub of a vehicle in FelGAN; and the “discriminator”, or another algorithm called a competitor, will distinguish a Collection includes real wheel photos and generator created images. This process will be repeated until the end of training.

Both algorithms are able to learn from their mistakes and improve continuously. After enough rounds of training, the images produced by the “generator” are real enough to deceive the human eye, or at least indistinguishable from real photos.

Audi experts use high-tech milling machines to make virtual wheel design a reality

FelGAN has an intuitive user interface based on Streamlit technology, which can effectively shorten the design cycle and improve the feedback efficiency between design and IT teams. In order for designers to use this software solution without relying on high-performance local hardware, computationally intensive AI application components will run in the cloud.

Humans and AI join forces

Another benefit of FelGAN is that the software assigns a numerical value to each design made by the AI. These values ​​are called “DNA” by the developers and can be used to reproduce the design at any time. In addition, Audi designers can enter their own designs and pictures into the program and add them to the virtual experiment interface, because specific algorithms also assign corresponding DNA values ​​to the images entered by the designers.

Audi experts use high-tech milling machines to make virtual wheel design a reality

In many cases, designers use FelGAN just to polish individual elements so that they can blend harmoniously with the overall design. At this time, in addition to the flexible application of this tool, personal creative vision and professional experience will play a decisive role. Ultimately, Audi experts use high-tech milling machines to create prototypes of the wheels in plastic or aluminum, making the virtual design a reality.

Innovations belonging to the world of Audi

The development and application of FelGAN is completely completed by the cooperation of Audi’s internal IT department and design department, which vividly demonstrates Audi’s professional level in the software field and the future artificial intelligence world.

Designers and developers view tires designed with FelGAN assistance

Thomas Knispel, head of Audi Machine Learning and Data Science, said: “In modern society, data brings immeasurable added value to the company and its employees. Audi is committed to becoming a data-driven company. To achieve this goal, we will Many departments use AI, and our data team will continue to explore new technologies.”

In the future, the technology behind FelGAN can also be extended to form a comprehensive AI design platform, providing inspiration for designers in other departments of Audi. In addition, Audi is developing an AI scoring system for evaluating the carbon balance of wheel designs generated by FelGAN. This once again shows that FelGAN is an important step for Audi on the road to becoming a data-driven digital company.