SimuLearn: Fast and Accurate Simulator to Support Morphing Materials Design and Workflows (accepted to UIST 2020)

Third Author, MFI funded, October 2018 - May 2020

Harnessing the active and transformative nature of materials, recent HCI advancements have created a library of shape-changing interfaces that afford new modalities of interaction and fabrication. 4D printing, in particular, empowers users to design and prototype more rapidly and economically. However, due to the lack of fast and accurate transformation predictions, currently available CAD tools cannot afford users to iterate designs that have complex topologies with high efficiency digitally. To address this issue, we take mesh-like structures as an example to introduce a novel SimuLearn system that combines finite element analysis (FEA) and graph convolutional networks (GCN) to truthfully (97% accuracy versus FEA) inform design decisions in real-time (0.6 seconds), and deploy our implementation in a computational design tool to unveil the enabled design space. Beyond 4D printing, SimuLearn also enriches the toolbox for shape-changing interfaces and paves the way for interactive CAD tools to unfold.

Paper link: https://drive.google.com/file/d/1rql3a9VcvxxsnEV3jQ4yFqH3-k4bb89c/view?usp=sharing