Artificial Intelligence (AI) has received significant attention in the field of the design of surface textures due to the excellent ability to analyze a large amount of data and thus reveal patterns between some complex phenomena. This paper reviews the main classifications of AI-aided surface texture design, including data-driven, model-driven, and data and model hybrid approaches. Data-driven approaches leverage large-scale datasets to extract effective design features via machine learning algorithms. These features are then utilized to optimize surface textures, ensuring they meet specific functional requirements. The model-driven approach is based on physical models and combines AI technology to perform parameter optimization and simulation to ensure the physical rationality of the design. By combining the advantages of data-driven and model-driven approaches, the data and model hybrid approach achieves a more efficient and accurate design process. In addition, the design of AI-aided surface textures for tribology, fluid dynamics and drag reduction, and biomedical applications is presented. Finally, a perspective on the current challenges as well as future research directions is presented.

Experiments have shown that surface texturing can enhance the tribological properties of bearings and seals made of soft materials such as polymers. However, the underlying mechanisms are still not fully understood and lack theoretical research. In this study, a mixed elastohydrodynamic model is specifically developed to explore the tribological behaviors of textured soft sliders. This model couples the flow, cavitation, deformation, and contact equations, considering both normal and shear stress effects with the fast Fourier transform method. The local deformation, contact and hydrodynamic pressure distributions of soft surfaces with multi-dimples are obtained, and a parametric study for the circular, triangular and partial texturing is conducted. It is found that the local deformations in different dimples may have a synergistic effect on providing additional hydrodynamic load support, termed the “barrel-like equivalent surface effect”. The textures concentrate the contact pressure at the dimple edges but may reduce the average contact pressure.