Rapid Design of Fully Soft Deployable Structures Via Kirigami Cuts and Active Learning
Leixin Ma,Mrunmayi Mungekar,Vwani Roychowdhury,M.K. Jawed
DOI: https://doi.org/10.1002/admt.202301305
IF: 6.8
2024-01-27
Advanced Materials Technologies
Abstract:Soft deployable structures revolutionize robotics and manufacturing by assuming intricate 3‐D shapes, challenging conventional rigid deployables. Overcoming fabrication hurdles, the approach employs a planar method with strain mismatches and kirigami cuts for precise shape control. Additionally, a data‐driven physics‐guided framework, utilizing autoencoders, streamlines the inverse design problem. Demonstrated through various shapes, the rapid design procedure aligns experiments with numerical predictions. Soft deployable structures – unlike conventional piecewise rigid deployables based on hinges and springs – can assume intricate 3‐D shapes, thereby enabling transformative soft robotic and manufacturing technologies. Their virtually infinite degrees of freedom allow precise control over the final shape. The same enabling high dimensionality, however, poses a challenge for solving the inverse problem: fabrication of desired 3D structures requires manufacturing technologies with extensive local actuation and control, and a trial‐and‐error search over a large design space. Both of these shortcomings are addressed by first developing a simplified planar fabrication approach that combines two ingredients: strain mismatch between two layers of a composite shell and kirigami cuts that relieves localized stress. In principle, it is possible to generate targeted 3‐D shapes by designing the appropriate kirigami cuts and the amount of prestretch (without any local control). Second, a data‐driven physics‐guided framework is formulated that reduces the dimensionality of the inverse design problem using autoencoders and efficiently searches through the "latent" parameter space in an active learning approach. The rapid design procedure is demonstrated via a range of target shapes, such as peanuts, pringles, flowers, and pyramids. Experiments and our numerical predictions are found to be in good agreement.
materials science, multidisciplinary