Dec. 5, 2025

We are pleased to announce the availability of a new AI-assisted sample centering capability at both CMCF beamlines. This update introduces fast and reliable automated sample centering to support high-throughput data collection within MxDC.

The new system is based on a Convolutional Neural Network model trained on thousands of sample images to identify pins and loops.  The system evaluates the sample images in real time and autonomously centers the target based on the identified objects in the video frame. A complete loop centering cycle takes approximately eight seconds to complete and eliminates much of the manual intervention previously required.

Initial operations indicate the method is working well, delivering robust and consistent sample positioning. The feature is an essential component of a fully automated, unattended data-collection access mode currently being implemented at both beamlines.