Sustainable design of new energy vehicle forms based on visual attention sequences and the Kolmogorov-Arnold Transformer
Xinhui Kang, Ziteng Zhao, Zimo Chen · Journal of Engineering Design · 2026
Against the backdrop of the global green transition and the booming new energy vehicle (NEV) market, the new energy sport utility vehicle (NEV-SUV) has become mainstream for its environmental attributes and performance advantages. The emotional experience conveyed by its appearance has become a key factor influencing purchase decisions. To achieve sustainable NEV-SUV form design, this study proposes an innovative approach integrating visual focus sequences with Kolmogorov-Arnold Transformer (KAT). Eye-tracking captures users’ focus points, then a morphological decomposition table of the NEV-SUV’s appearance is constructed. Core emotional vocabulary is collected and refined using a large-scale language model, followed by calculation of vocabulary weights via an improved game-theoretic method. The top three words are selected, and KAT establishes a nonlinear mapping between NEV-SUV focus features and emotional vocabulary, generating optimal design parameter combinations. Finally, Rhino is used for 3D modelling, with generative AI for fast rendering and point cloud testing to verify accuracy. Results show the final design received high user ratings, and compared with traditional neural networks, KAT better captured emotional features, with advantages in speed, accuracy, and robustness. The findings provide scientific data support for NEV-SUV appearance design, enhancing competitiveness and promoting sustainable development in the NEV industry.