Leveraging Generative AI and CAD Automation for Efficient Automotive Wheel Design with Limited Data

Abstract This study developed a method that leverages generative artificial intelligence and automated computer-aided design (CAD) technologies for the design of automotive wheel rims. First, numerous two-dimensional (2D) wheel rim designs are automatically generated by inputting prompts into Stable Diffusion, an image generation model, without predefining design parameters and constraints. Subsequently, the reasonableness of the generated 2D designs is examined, and reasonable designs are transformed into three-dimensional CAD models. The proposed method completes model training with a small amount of data and simultaneously establishes a mechanism for verifying the reasonableness of data generation, achieving reliable wheel rim design generation. Compared to existing big data methods, this approach is more easily applicable to industrial applications.