Most monolithic engineering materials have positive Poisson’s ratio and contract laterally relative to the direction of imparted strain. However, auxetics are a class of materials that expand laterally to the direction in which they are strained, exhibiting negative Poisson’s ratio (NPR). The key to an auxetic material’s NPR is the internal structural mechanism that creates a rotational or leveraging effect of unit cell struts to the surrounding cells causing counterintuitive expansion. Combining an array of auxetic cells forms metamaterials with tunable gradient properties depending on the unit cell geometry used and concentrations of the respective Poisson’s ratio values. Many unit cell architectures have been studied; however, significant focus has been conducted on the variations of the reentrant honeycomb cell which behaves within a range of strain, as an effective NPR unit cell expanding positively in all directions with positive imparted strain. Reentrant honeycomb cells are the focus of this study because of their large basis of research and relative geometric simplicity for tuning behavior. Due to the unique mechanical properties of auxetics, NPR materials are used in engineering applications, especially when thematerials undergo significant deformation. However, the behavior of auxetics in soft-bodied materials is still being developed and makes their interaction in applications difficult to predict. Applications of auxetic metamaterials have been explored in previous literature, ranging from impact lattice structures, medical stents, and others described in depth in previous review studies, whose main function is derived from their behavior under strain. Of particular importance to our work is the application of auxetic metamaterials in 2D and 3D soft bodies due to this counterintuitive deformation behavior. 2D elastomeric lattices are currently in use and being explored in sportswear and biomedical films that are able to conform to the body. However, out-of-plane strain is still a design fault, impeding progress for both applications. 3D application studies have also been conducted on the use of biomimetic soft-body manipulators using fluidic control. Manipulator applications utilize conventional kinematic models for simple movements but such models restrain researchers to a predefined control methodology when creating more complex control systems. Researchers have explored auxetic geometries to create a soft-bodied cylinder using auxetic cell buckling failure for angular actuation. This work demonstrated fundamental aspects of NPR in 3D actuation and incentivized our study to develop a workflow for designing dynamic auxetics, using a similar form of an angularly actuated cylinder. Aside from the pressure-driven actuation control system, in many of these cases, the only way for the intricate structures to bemanufactured is through detailed casting or, as we use in our approach, via an additive manufacturing system. Additive manufacturing is an ideal method of creating various types of metamaterials because 3D printers are capable of fabricating complex architectures. The coupled ability to rapidly and iteratively discover new designs with machine learning and experimentally validate results with additive manufacturing marks a novel discovery process in manufacturing and materials characterization. The programmability of the additive J. K. Wilt, C. Yang, Prof. G. X. Gu Department of Mechanical Engineering University of California, Berkeley 6177 Etcheverry Hall, Berkeley, CA 94720-1740, USA E-mail: ggu@berkeley.edu

中文翻译：

---

通过深度学习加速拉胀超材料设计

大多数整体工程材料具有正泊松比并相对于施加的应变方向横向收缩。然而，拉胀材料是一类横向膨胀到应变方向的材料，表现出负泊松比 (NPR)。拉胀材料 NPR 的关键是内部结构机制，它会产生晶胞支柱对周围细胞的旋转或杠杆作用，从而导致违反直觉的膨胀。根据所使用的晶胞几何形状和各自泊松比值的浓度，组合一系列拉胀细胞形成具有可调梯度特性的超材料。已经研究了许多单元结构；然而，重点关注在一定应变范围内表现的可重入蜂窝单元的变化，作为有效的 NPR 单元单元，在所有方向上正向扩展，并具有正传递的应变。重入蜂窝单元是本研究的重点，因为它们具有广泛的研究基础和调整行为的相对几何简单性。由于拉胀材料独特的机械性能，NPR材料被用于工程应用，特别是当材料发生显着变形时。然而，软体材料中拉胀的行为仍在开发中，这使得它们在应用中的相互作用难以预测。拉胀超材料的应用已经在之前的文献中进行了探索，包括冲击晶格结构、医疗支架、和其他人在之前的评论研究中进行了深入描述，其主要功能来自他们在压力下的行为。由于这种违反直觉的变形行为，我们的工作特别重要的是拉胀超材料在 2D 和 3D 软体中的应用。2D 弹性晶格目前正在使用，并且正在探索能够贴合身体的运动服和生物医学薄膜。然而，面外应变仍然是一个设计错误，阻碍了这两种应用的进展。还对使用流体控制的仿生软体机械手的使用进行了 3D 应用研究。机械手应用程序利用传统运动学模型进行简单运动，但此类模型在创建更复杂的控制系统时限制研究人员使用预定义的控制方法。研究人员已经探索了拉胀几何形状，以使用拉胀单元屈曲失效进行角驱动来创建软体圆柱体。这项工作展示了 NPR 在 3D 驱动中的基本方面，并激励我们的研究开发用于设计动态拉胀的工作流程，使用类似形式的角驱动气缸。除了压力驱动的驱动控制系统外，在许多情况下，制造复杂结构的唯一方法是通过详细的铸造，或者，正如我们在我们的方法中所使用的，通过增材制造系统。增材制造是创建各种类型超材料的理想方法，因为 3D 打印机能够制造复杂的架构。通过机器学习快速和迭代地发现新设计并通过增材制造对结果进行实验验证的耦合能力标志着制造和材料表征方面的新发现过程。添加剂的可编程性 JK Wilt, C. Yang, Prof. GX Gu 加州大学伯克利分校机械工程系 6177 Etcheverry Hall, Berkeley, CA 94720-1740, USA 电子邮件：ggu@berkeley.edu