Session: S-07 Computing Metamaterials

Paper Number: 170856

170856 - Physical Networks in Metamaterials With Neural Networks 

The implementation of intelligence in mechanical systems is essential for advancing autonomous systems. Such systems must actively adapt to environments and demonstrate capabilities of learning, memory, and decision-making. The reliance on either solely mechanical platforms or computational resources limits overall performance of intelligence. The intelligent mechanical systems can be developed by combining adaptable materials with computational algorithms such as machine learning and deep learning. In this context, this research introduces a novel physical intelligent system—a Physical Network (PN)—that integrates metamaterials with neural networks.

Our approach addresses these challenges through the development of physical intelligence (PI), which functions as an intelligent body interacting with and adapting to environmental stimuli, complementing computational intelligence (CI) which serves as the brain. The proposed system employs mechanical metamaterials with adaptable properties, utilizing a single central accelerometer as the sole electronic sensor. When external forces are applied to the metamaterial platform, mechanical wave propagation is detected and analyzed. Neural networks are trained to classify interaction sources or estimate properties of external objects based on these measurements.

Concurrently, the metamaterial structure undergoes optimization to enhance neural network performance through a two-tier framework: design optimization (high-level) and neural network training (low-level). The metamaterials are well known as their programmability and adaptability. Previous research proposed the effectiveness of inverse design of metamaterials to achieve target properties (e.g. maximizing energy absorption) using differentiable simulations of metamaterials. Inspired by the efficiency of differentiable simulation to optimize the designs, we implement differentiable simulation of metamaterials to obtain gradient information with respect to design parameters. Since the high-level optimization for design includes the iterative computational graph of training neural networks, the implicit differentiation is employed to overcome computational barriers in backpropagation. The gradient-based approach for metamaterial design optimization is benchmarked against evolutionary strategies to explore global optima despite efficiency trade-offs. This co-optimization process represents a form of robotic evolution that enhances environmental sensing capabilities.

The experimental platform consists of metamaterials fixed at the base with continuous force functions applied to the top surface, while measurements are obtained from a single accelerometer on the central block. As environmental complexity increases, our experimental results demonstrate that structural design modifications improve algorithmic performance. This interaction between computational algorithms and physical evolution aligns with evolutionary processes observed in nature.

This research presents an active physical system that evolves in conjunction with neural networks, streamlining interactions between environment, body dynamics, and information processing. The proposed approach advances the field toward genuine robotic autonomy with minimal electronic sensing requirements, enabling detection of environmental stimuli and supporting decision-making in autonomous systems.

Presenting Author: Kyungmi Na Georgia Institute of Technology

Presenting Author Biography: Kyungmi Na is currently a PhD student in Machine Learning program from Georgia Institute of Technology. She received a master degree in Aerospace Engineering from Korea Advanced Institute of Technology. She earned(received) a bachelor's degree in mechanical engineering from Pohang University of Science and Technology. Her interests are scientific machine learning for simulation and design optimization of metamaterials.