Abstract: To satisfy the participation of the public with different disciplines in the design process of high-speed rails, we conduct a research on computer-aided conceptual design of high-speed rails based on human-AI collaboration. This work aims to lower the barrier of designing through human-machine co-creation and collaboration, inspire the creativity and imagination of designers, visualize their design concepts that flash in mind, and simplify the time and efforts while creating detailed sketches. First, based on creative support theory and the different stages in the conceptual design of high-speed rails, we propose three human-AI collaborative ideation strategies to promote divergent thinking and convergent thinking. Next, we create a high-quality dataset of high-speed rails using the deep generative adversarial neural network (GAN) to implement the proposed creative ideation strategies. The system takes the designer's input as a limitation to generate both continuity and novelty designs in line with the aesthetics. Finally, we invite participants to verify the validity of our method and report the results of the qualitative analysis, aiming to provide a reference for future research on the collaborative design of crowd intelligence.

Key words: crowd design, human-ai co-creation, computer-aided design, generative design, GAN