Harnessing Interpretable Machine Learning for Origami Feature Design and Pattern Selection
share
Engineering design of origami systems is challenging because comparing different origami patterns requires using categorical features and evaluating multi-physics behavior targets introduces multi-objective problems. This work shows that a decision tree machine learning method is particularly suitable for the inverse design of origami. This interpretable machine learning method can reveal complex interactions between categorical features and continuous features for comparing different origami patterns, can tackle multi-objective problems for designing active origami with multi-physics performance targets, and can extend existing origami shape fitting algorithms to further consider non-geometrical performances of origami systems. The proposed framework shows a holistic way of designing active origami systems for various applications such as metamaterials, deployable structures, soft robots, biomedical devices, and many more.
READ FULL TEXT