Recent advances in generative AI have enabled a variety of applications across different industries. Particularly in manufacturing, where surrogate models and 3D shape generation are being explored as a means of shortening development cycles. However, a few studies have addressed the issue of generating conditional shapes using partial geometry under limited datasets. This study proposes a novel, image-based method for generating the internal structures of automobiles, for which thin sheet metal components are widely used, based on 3D boundary shapes. The outer boundary is represented using B-spline curves and the internal structure is formulated as a height difference in the z direction from a reference surface, which is defined by B-spline boundary curves. This differential representation ensures that the edges of the generated image always satisfy the boundary conditions, even with limited training data. This method was applied to an inner panel of a vehicle hood, and the generated shape could be easily integrated into the original model. This demonstrates the effectiveness and practicality of the proposed approach.
SUGIURA et al. (Wed,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: