Metamaterials that Learn to Design other Metamaterials
Abstract
The objective of this research is to create a new paradigm for designing multi-functional materials by using active metamaterials that first mechanically learn desired behaviors via real- time physical optimization and then transfer what was learned to a passive easily fabricated- version of the design. The real-time physical optimization approach mechanically embodies how artificial neural networks mathematically learn to map inputs to desired outputs. To create the new design paradigm, we will first leverage simulation and experimentation to learn what aspects of active metamaterials affect their mechanical learning capabilities and why. We will then use this new knowledge to guide the design, fabrication, and demonstration of an active micro-scale metamaterial that applies mechanical learning to design a passive metamaterial that achieves a desired set of bulk behaviors. The proposed research will produce a greater understanding of the fundamental principles that govern how the process of learning can be achieved mechanically, which, if successfully understood, could uncover some of nature’s secrets relating to how real brains leverage physical matter to enable or enhance learning. Moreover, this research will produce the first multi-functional material that has ever been designed using another material, and the material will be designed without human intervention or insight. The proposed design approach will enable the rapid design of multi-functional materials of critical importance to the US air force while immediately and simultaneously considering all the relevant effects experienced by the actual material (e.g., nonlinearities, hysteresis, friction, dynamic behaviors, fabrication tolerances and errors, and viscoelastic, plastic, or other complex mechanical or thermal effects). And, it will do so without requiring expensive and time-consuming computation required by other design approaches that don’t leverage the real-time computational power of natural law.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 07, 2023
- Source ID
- FA95502210008
Entities
People
- Jonathan B. Hopkins
Organizations
- Air Force Office of Scientific Research
- United States Air Force
- University of California, Los Angeles