A Design Method for Automotive Front Styling Based on AIGC and Kansei Engineering

This study proposes a novel automotive front styling design method that integrates Retrieval-Augmented Generation Large Language Models (RAG-LLM) with Kansei Engineering and validates its effectiveness through empirical research. The research is divided into five phases: building the LLM platform, constructing the RAG-LLM platform, conducting knowledge-based question answering testing, evaluating design capabilities, and testing image generation. The test results show that the RAG-LLM model significantly outperforms traditional LLM models in generating Kansei vocabulary, as well as in the accuracy and completeness of design proposals. Additionally, when design proposals generated by RAG-LLM were input into image generation AI, the resulting automotive design images demonstrated improvements in design quality, emotional intent, and consumer preferences. This study also demonstrates that the RAG-LLM framework effectively optimizes the application of Kansei Engineering in automotive styling design, enhancing both design efficiency and quality.