Department of Mechanical Engineering and Applied Mechanics, University of Pennsylvania, Philadelphia, PA, USA.
Materials and Manufacturing Directorate, Air Force Research Laboratory, Wright-Patterson Air Force Base, OH, USA.
Nature. 2021 Oct;598(7879):39-48. doi: 10.1038/s41586-021-03623-y. Epub 2021 Oct 6.
Mechanical mechanisms have been used to process information for millennia, with famous examples ranging from the Antikythera mechanism of the Ancient Greeks to the analytical machines of Charles Babbage. More recently, electronic forms of computation and information processing have overtaken these mechanical forms, owing to better potential for miniaturization and integration. However, several unconventional computing approaches have recently been introduced, which blend ideas of information processing, materials science and robotics. This has raised the possibility of new mechanical computing systems that augment traditional electronic computing by interacting with and adapting to their environment. Here we discuss the use of mechanical mechanisms, and associated nonlinearities, as a means of processing information, with a view towards a framework in which adaptable materials and structures act as a distributed information processing network, even enabling information processing to be viewed as a material property, alongside traditional material properties such as strength and stiffness. We focus on approaches to abstract digital logic in mechanical systems, discuss how these systems differ from traditional electronic computing, and highlight the challenges and opportunities that they present.
机械机制已经被用于处理信息几千年了,著名的例子包括古希腊的安提基西拉机械和查尔斯·巴贝奇的分析机。更近一些,由于更好的微型化和集成潜力,电子形式的计算和信息处理已经超越了这些机械形式。然而,最近已经引入了几种非常规的计算方法,这些方法融合了信息处理、材料科学和机器人技术的思想。这增加了新型机械计算系统的可能性,这些系统通过与环境相互作用和适应来增强传统的电子计算。在这里,我们讨论了将机械机制及其相关的非线性作为信息处理的一种手段,着眼于一个框架,在这个框架中,适应性材料和结构充当分布式信息处理网络,甚至可以将信息处理视为一种材料特性,与传统的材料特性(如强度和刚度)并列。我们专注于机械系统中抽象数字逻辑的方法,讨论这些系统与传统电子计算的不同之处,并强调它们带来的挑战和机遇。