Through the APEX and ARMOR projects, the national laboratory is having staff control the equipment and experimental parameters. The analysis of 3D-printed images has been reduced from several minutes to a few milliseconds.
Lawrence Livermore National Laboratory details how it combines artificial intelligence, robotics and automation to accelerate advanced manufacturing, materials discovery and experimental science, in conjunction with the Department of Energy and the National Nuclear Security Administration.
The distinction the lab insists on making separates two levels. An automated lab performs pre-programmed tasks; an autonomous lab uses AI to collect data, analyze it, and determine what needs to happen next. In this second model, agents don't just read the results, they control the hardware and adjust the experimental parameters in real time. Two projects illustrate the approach: APEX (Alloy Prediction and Experimentation), which combines AI, robotics, and automated workflows for alloy development, and ARMOR, which explores robotic execution in an unstructured environment.
The payoff is sometimes very concrete: applied to process images from 3D printing, machine learning reduces an analysis that used to require several minutes of manual review per image to a few milliseconds. For the lab and its NNSA partners, the ultimate goal is a closed loop linking AI-assisted design, high-performance computing, advanced manufacturing, and feedback.
Read the full article on the original source
Go to the source →