A physical adaptive material motor unit neural network: a hygromorph composite material machine
Researchers have developed a novel neural network that utilizes adaptive materials, specifically wood and carbon black composites, to create intelligent machines. These machines are sensitive to temperature and humidity, mimicking muscle contraction to perform tasks like dynamic shading control in buildings. The system is trained on over 350 experimental data points using a new data-aware backpropagation method, allowing it to learn and predict behavior as more data becomes available. The material motor unit neural network has demonstrated its ability to optimize configurations for consistent shading outputs under varying environmental conditions. AI
IMPACT Introduces a novel approach to embedding learning capabilities directly into materials for autonomous machine functions.