An exploratory study on “cross-inference method” for supporting idea generation process of industrial designers

In the early stages of an industrial design process, existing concepts can be obstacles for generating fresh ideas. A computer assisted method to find a way out from this situation is proposed. Firstly, two groups of different type products (products “A” and “B”) should be chosen, with the consideration that they have some kind of common components which realize their functions. Then the patterns of relationship between design concepts and elements of form of type “A” products are learned in a neural network to make a model to infer form of A (named A model). The same process is conducted for type “B” products to make B model. Subsequently, for obtaining a new pattern of the elements of form, put a concept pattern of the product “A” into the B model, and vice versa. Then an unforeseen concept form relation, leading to a feasible new idea, can be got by this method. A case study in a student project are shown