4D printing technology allows for complex material distributions in voxelized structures, offering enormous design flexibility for shape-transforming materials. However, the tremendous design space due to numerous voxels also poses a challenge in efficiently finding appropriate designs to achieve target shape changes.
This recent work presents a novel machine learning (ML) and evolutionary algorithm (EA) based approach to guide the voxel-based design for 4D printed active composite beams. After being trained by using the data generated by finite element simulations, ML can predict shape-change accurately and rapidly for the given material-distribution designs. EA empowered with ML enables the efficient search for optimal designs to achieve complicated shape changes. Further integrating the ML-EA with computer vision algorithms permits a new streamlined design-fabrication paradigm that allows the quick fabrication of 4D-printed beams with shape changes based on the conceptual hand-drawn lines. The proposed 4D-printing design paradigm can be extended to various material systems with different environmental responsive mechanisms.
This work is conducted by Prof. Qi’s group at Georgia Tech in collaboration with Prof. Kun Zhou of Nanyang Technological Unviersity, Prof. Frédéric Demoly of Univ. Bourgogne Franche-Comté, and Prof. Ruike Zhao of Stanford University. The first author is Dr. Xiaohao Sun.
Sun, X., Yue, L., Yu, L., Shao, H., Peng, X., Zhou, K., Demoly, F., Zhao, R., Qi, H. J., Machine Learning-Evolutionary Algorithm Enabled Design for 4D-Printed Active Composite Structures. Adv. Funct. Mater. 2021, 2109805. https://doi.org/10.1002/adfm.202109805