skip to main content

Title: Harnessing interpretable machine learning for holistic inverse design of origami
Abstract

This work harnesses interpretable machine learning methods to address the challenging inverse design problem of origami-inspired systems. We established a work flow based on decision tree-random forest method to fit origami databases, containing both design features and functional performance, and to generate human-understandable decision rules for the inverse design of functional origami. First, the tree method is unique because it can handle complex interactions between categorical features and continuous features, allowing it to compare different origami patterns for a design. Second, this interpretable method can tackle multi-objective problems for designing functional origami with multiple and multi-physical performance targets. Finally, the method can extend existing shape-fitting algorithms for origami to consider non-geometrical performance. The proposed framework enables holistic inverse design of origami, considering both shape and function, to build novel reconfigurable structures for various applications such as metamaterials, deployable structures, soft robots, biomedical devices, and many more.

Authors:
;
Publication Date:
NSF-PAR ID:
10379887
Journal Name:
Scientific Reports
Volume:
12
Issue:
1
ISSN:
2045-2322
Publisher:
Nature Publishing Group
Sponsoring Org:
National Science Foundation
More Like this
  1. Self-folding origami, structures that are engineered flat to fold into targeted, three-dimensional shapes, have many potential engineering applications. Though significant effort in recent years has been devoted to designing fold patterns that can achieve a variety of target shapes, recent work has also made clear that many origami structures exhibit multiple folding pathways, with a proliferation of geometric folding pathways as the origami structure becomes complex. The competition between these pathways can lead to structures that are programmed for one shape, yet fold incorrectly. To disentangle the features that lead to misfolding, we introduce a model of self-folding origami that accounts for the finite stretching rigidity of the origami faces and allows the computation of energy landscapes that lead to misfolding. We find that, in addition to the geometrical features of the origami, the finite elasticity of the nearly-flat origami configurations regulates the proliferation of potential misfolded states through a series of saddle-node bifurcations. We apply our model to one of the most common origami motifs, the symmetric “bird's foot,” a single vertex with four folds. We show that though even a small error in programmed fold angles induces metastability in rigid origami, elasticity allows one to tune resilience tomore »misfolding. In a more complex design, the “Randlett flapping bird,” which has thousands of potential competing states, we further show that the number of actual observed minima is strongly determined by the structure's elasticity. In general, we show that elastic origami with both stiffer folds and less bendable faces self-folds better.« less
  2. Abstract

    Controlling the deformation of a soft body has potential applications in fields requiring precise control over the shape of the body. Areas such as medical robotics can use the shape control of soft robots to repair aneurysms in humans, deliver medicines within the body, among other applications. However, given known external loading, it is usually not possible to deform a soft body into arbitrary shapes if it is fabricated using only a single material. In this work, we propose a new physics-based method for the computational design of soft hyperelastic bodies to address this problem. The method takes as input an undeformed shape of a body, a specified external load, and a user desired final shape. It then solves an inverse problem in design using nonlinear optimization subject to physics constraints. The nonlinear program is solved using a gradient-based interior-point method. Analytical gradients are computed for efficiency. The method outputs fields of material properties which can be used to fabricate a soft body. A body fabricated to match this material field is expected to deform into a user-desired shape, given the same external loading input. Two regularizers are used to ascribe a priori characteristics of smoothness and contrast, respectively,more »to the spatial distribution of material fields. The performance of the method is tested on three example cases in silico.

    « less
  3. Modern machine learning has achieved impressive prediction performance, but often sacrifices interpretability, a critical consideration in many problems. Here, we propose Fast Interpretable Greedy-Tree Sums (FIGS), an algorithm for fitting concise rule-based models. Specifically, FIGS generalizes the CART algorithm to simultaneously grow a flexible number of trees in a summation. The total number of splits across all the trees can be restricted by a pre-specified threshold, thereby keeping both the size and number of its trees under control. When both are small, the fitted tree-sum can be easily visualized and written out by hand, making it highly interpretable. A partially oracle theoretical result hints at the potential for FIGS to overcome a key weakness of single-tree models by disentangling additive components of generative additive models, thereby reducing redundancy from repeated splits on the same feature. Furthermore, given oracle access to optimal tree structures, we obtain l2 generalization bounds for such generative models in the case of C component functions, matching known minimax rates in some cases. Extensive experiments across a wide array of real-world datasets show that FIGS achieves state-of-the-art prediction performance (among all popular rule-based methods) when restricted to just a few splits (e.g. less than 20). We findmore »empirically that FIGS is able to avoid repeated splits, and often provides more concise decision rules than fitted decision trees, without sacrificing predictive performance. All code and models are released in a full-fledged package on Github.« less
  4. Magnetoelastic thin shells exhibit great potential in realizing versatile functionalities through a broad range of combination of material stiffness, remnant magnetization intensity, and external magnetic stimuli. In this paper, we propose a novel computational method for forward simulation and inverse design of magnetoelastic thin shells. Our system consists of two key components of forward simulation and backward optimization. On the simulation side, we have developed a new continuum mechanics model based on the Kirchhoff-Love thin-shell model to characterize the behaviors of a megnetolelastic thin shell under external magnetic stimuli. Based on this model, we proposed an implicit numerical simulator facilitated by the magnetic energy Hessian to treat the elastic and magnetic stresses within a unified framework, which is versatile to incorporation with other thin shell models. On the optimization side, we have devised a new differentiable simulation framework equipped with an efficient adjoint formula to accommodate various PDE-constraint, inverse design problems of magnetoelastic thin-shell structures, in both static and dynamic settings. It also encompasses applications of magnetoelastic soft robots, functional Origami, artworks, and meta-material designs. We demonstrate the efficacy of our framework by designing and simulating a broad array of magnetoelastic thin-shell objects that manifest complicated interactions between magnetic fields,more »materials, and control policies.« less
  5. Abstract The Kresling truss structure, derived from Kresling origami, has been widely studied for its bi-stability and various other properties that are useful for diverse engineering applications. The stable states of Kresling trusses are governed by their geometry and elastic response, which involves a limited design space that has been well explored in previous studies. In this work, we present a magneto-Kresling truss design that involves embedding nodal magnets in the structure, which results in a more complex energy landscape, and consequently, greater tunability under mechanical deformation. We explore this energy landscape first along the zero-torque folding path and then release the restraint on the path to explore the complete two-degree-of-freedom behavior for various structural geometries and magnet strengths. We show that the magnetic interaction could alter the potential energy landscape by either changing the stable configuration, adjusting the energy well depth, or both. Energy wells with different minima endow this magneto-elastic structure with an outstanding energy storage capacity. More interestingly, proper design of the magneto-Kresling truss system yields a tri-stable structure, which is not possible in the absence of magnets. We also demonstrate various loading paths that can induce desired conformational changes of the structure. The proposed magneto-Kresling trussmore »design sets the stage for fabricating tunable, scalable magneto-elastic multi-stable systems that can be easily utilized for applications in energy harvesting, storage, vibration control, as well as active structures with shape-shifting capability.« less