Human-Machine Co-Ideation Via Combinational Generative Model

Abstract

Ideation is a crucial step in the engineering design process. Designers use the ideation process to create new concepts and prototypes that are creative and innovative. The current ideation workflow requires that designers produce new designs in accordance with the requirements of the product, primarily relying on the designers’ personal expertise and experiences. To push the boundaries of human-machine co-design and help designers with the idea-generation process, this paper proposes an integrated approach by combining generative adversarial networks (GANs) and data mining techniques. The proposed approach consists of an image encoder module and a cross-domain object combination decoder module. The image encoder module encodes the image structure information into latent space, and the cross-domain object combination decoder module combines object images together according to the user’s preference using GANs, which generate new design images. A design case study is used to evaluate the new ideation approach and reveal not only strong cross-domain concept combination capabilities but also improvement in designers’ workflow and provision of novelty to the design case.

Keywords: Ideation, Combinational Creativity, Visual Stimuli, Artificial intelligence, Generative Adversarial Networks