Transforming Car Manufacturing: How Generative AI and Machine Learning Are Revolutionizing Production
DOI:
https://doi.org/10.70179/vjy6gc79Keywords:
Generative Adversarial Networks (GANs), Deep Learning, Manufacturing Automation, Data-Confirmed Neural Networks, High-Resolution Surface Generation, Automotive Body Shop, Latent Vector Transformation, Face Generator Metaphor, Production-Oriented Applications, Label-to-Attribute SetupAbstract
The Daimler AG trains Generative Adversarial Network (GAN) to grow the data-confirmed neural networks. The ultimate goal is a production-oriented application that enables faster manufacturing, improved quality, and a more efficient use of resources through a higher level of automation. The success of deep learning-based methods, particularly Generative Adversarial Networks (GANs), has led to new opportunities in generative AI in the manufacturing domain. In a proof of concept, we use a face generator as a metaphor for the generation of complex workpiece surfaces. The GAN transforms a low-resolution surface map - the so-called latent vector - into a high-resolution area on the generated face surface. This is used for the manufacturing of a corresponding face in an automotive body shop. The method is based on a label-to-attribute setup and includes a pre-fitting step for control point generation.
