Lawrence Livermore researchers built a camera system that watches 3D prints layer by layer, catching defects before they ruin a part.

Watching prints in real time

Scientists at Lawrence Livermore National Laboratory have built a camera-based system that monitors 3D-printed parts as they print. The setup uses machine learning to measure strand diameter, spacing, and other features layer by layer.

The system targets direct ink writing, where soft or paste-like materials are deposited through a nozzle. Small gaps or broken strands in these prints can ruin mechanical performance, but catching those problems used to mean waiting until the print finished, then running expensive X-ray or mechanical tests.

LLNL's approach changes that timeline. A camera captures images as each layer is deposited. Software identifies the newest strands and calculates measurements in milliseconds. In tests across 55 parts, the automated system matched human measurements within a few micrometers.

Speed and scale

The math on speed is hard to ignore. A person manually measuring a large image might spend 20 minutes to an hour on the task. The automated pipeline finishes in milliseconds. That is roughly 100,000 times faster on average.

The researchers also demonstrated the system on a 25 by 25 centimeter cushion, collecting about 2,500 images from a single layer. The resulting spatial map revealed filament diameters changing across the print, pointing to a slight platform tilt that an average measurement would have missed.

What comes next

Brian Weston, the project's technical lead, said the team now has a reusable starting point that can adapt to new cameras and part geometries with minimal retraining. The capability is expected to move to the Kansas City National Security Complex for evaluation on production-relevant systems.

Longer term, the inspection data could feed digital twins that link measured structure to predicted performance. That would let a printer not just detect defects, but decide whether a part still meets requirements or needs to be scrapped. When the machine can inspect its own work, the line between printing and quality control disappears.

Disclosure: Some links are affiliate links. We may earn a small commission at no extra cost to you.

Comments (0)

No comments yet. Be the first!

Leave a Comment