Sign In

Communications of the ACM

ACM TechNews

AI-Powered Material Can Learn Behaviors, Adapt

View as: Print Mobile App Share:

A portion of the mechanical neural network.

Credit: Flexible Research Group/UCLA

Mechanical engineers at the University of California, Los Angeles have developed an artificial intelligence-powered material that learns behaviors over time, and can adjust to changing circumstances.

The so-called mechanical neural network (MNN) features a structural system of independently tunable beams arranged in a triangular lattice pattern.

The researchers said each beam consists of a "voice coil, strain gauges, and flexures that enable the beam to change its length, adapt to its changing environment in real time, and interact with other beams in the system."

An optimization algorithm uses strain-gauge data to calculate rigidity values to govern the network's adaptation, determining how much force should be applied. Cameras on the MNN's outer nodes check the strain-gauge system's validity.

From Interesting Engineering
View Full Article


Abstracts Copyright © 2022 SmithBucklin, Washington, DC, USA


No entries found