DMREF/Collaborative Research: Design and Optimization of Granular Metamaterials using Artificial Evolution

DMREF/协作研究：利用人工进化设计和优化颗粒超材料

• 批准号: 2118810
• 负责人: Joshua Bongard
• 金额: $ 40.11万
• 依托单位: University of Vermont & State Agricultural College
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-09-01 至 2025-08-31
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2118810&HistoricalAwards=false
• 关键词: DMREF; Collaborative; Research; Design; Optimization

Metamaterials capable of accessing multiple properties will lead to systems possessing a wide range of functions. Such multifunctional metamaterials will increase the autonomy, efficiency, and lifespan of systems and structures by dynamically adapting to task demands or changes in the environment. Granular metamaterials are an advantageous platform for such dynamic programmability, as the grain properties can be widely tuned to achieve different responses. Granular metamaterial response is dependent on many variables, including the grain arrangement, grain mass, modulus, and shape, friction or other interactions between grains, as well as whether the walls of the container are held fixed or can move in response to forces from the grains. With such a vast number of possible combinations of micro-structural variables, the task of designing the complex relationship between micro-structure and bulk properties is daunting. This Designing Materials to Revolutionize and Engineer our Future (DMREF) research applies evolutionary algorithms to efficiently search the immense parameter space of granular metamaterial designs for specific material properties, as well as identifying how a design can be perturbed to actively transition from one set of desired bulk properties to another. The project will establish a new artificial intelligence-driven approach to the design and optimization of granular metamaterials with adaptable properties. These materials will aid US productivity and prosperity by providing additional means to find and use materials impacting robotics and other technical areas.Future granular metamaterials will exhibit increased dynamic plasticity, enabling responses to different environmental inputs or task demands by reconfiguring their physical structure. This project addresses two issues: (1) How can granular assemblies with specific desired material properties be automatically designed? (i.e., What should the grains' arrangement, moduli, shapes, masses, friction coefficients, and other grain-scale properties be to yield a given bulk material property?); and (2) How can a series of granular assemblies that allow continuous, time-ordered changes in material properties be automatically designed? (i.e., Which set of grain-scale variables should change and by how much? If multiple solutions are found, which ones can be realized in experiments most easily?) To address these issues, a closed-loop procedure will be developed and implemented wherein physics-based discrete element method (DEM) simulations, evolution-based optimization, and physical realizations are combined to produce granular metamaterials with desired, optimal, and adaptable material properties. New knowledge and tools that will be generated by this project include: (i) alternative evolutionary algorithms for designing granular metamaterials capable of changing their own configuration and properties; (ii) physics-based DEM simulations that can predict the properties of granular materials with active grains; and (iii) New synthesis strategies for multifunctional grains and grain assemblies.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

能够获得多种特性的超材料将导致具有广泛功能的系统。这种多功能超材料将通过动态适应任务需求或环境变化来提高系统和结构的自主性、效率和寿命。颗粒超材料是实现这种动态可编程性的有利平台，因为颗粒性质可以广泛调整以实现不同的响应。颗粒超材料的响应取决于许多变量，包括颗粒排列、颗粒质量、模量和形状、颗粒之间的摩擦或其他相互作用，以及容器壁是否固定或可以响应来自颗粒的力而移动。由于微观结构变量的可能组合如此之多，设计微观结构与体性能之间复杂关系的任务是艰巨的。这项设计材料以革新和工程我们的未来（DMREF）研究应用进化算法来有效地搜索颗粒超材料设计的巨大参数空间，以获得特定的材料特性，以及确定设计如何被扰动以积极地从一组所需的大块特性过渡到另一组。该项目将建立一种新的人工智能驱动的方法来设计和优化具有适应性的颗粒超材料。这些材料将为寻找和使用影响机器人和其他技术领域的材料提供额外的手段，从而有助于美国的生产力和繁荣。未来的颗粒状超材料将表现出更高的动态可塑性，能够通过重新配置其物理结构来响应不同的环境输入或任务需求。该项目解决了两个问题：(1)如何自动设计具有特定所需材料特性的颗粒组件？（即，晶粒的排列、模量、形状、质量、摩擦系数和其他晶粒尺度性质应该是什么，才能产生给定的块状材料性质？）；(2)如何自动设计一系列允许材料属性连续、有序变化的颗粒组件？(即，哪一组粒度变量应该改变，改变多少？如果有多种解决方案，哪一种最容易在实验中实现？)为了解决这些问题，将开发和实施一个闭环程序，其中基于物理的离散元素方法（DEM）模拟，基于进化的优化和物理实现相结合，以生产具有理想，最佳和适应性材料特性的颗粒超材料。该项目将产生的新知识和工具包括：(i)设计能够改变自身结构和特性的粒状超材料的替代进化算法；（ii）基于物理的DEM模拟，可以预测具有活性颗粒的颗粒材料的性质；(3)多功能晶粒和晶粒组合的新合成策略。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Universal Mechanical Polycomputation in Granular Matter

颗粒物质中的通用机械多计算

• DOI: 10.1145/3583131.3590520
• 发表时间: 2023
• 期刊: Proceedings of the Genetic and Evolutionary Computation Conference
• 作者: Parsa, Atoosa;Witthaus, Sven;Pashine, Nidhi;O'Hern, Corey;Kramer-Bottiglio, Rebecca;Bongard, Josh
• 通讯作者: Bongard, Josh