Automated alignment at APS (Nov 2026)

Title: AutoFocus: the AI-driven optics autoalignment software at the APS

Speaker: Luca Rebuffi (Advanced Photon Source, Argonne National Laboratory)

Date: Wednesday, November 18th, 2026 – 11am

Location: Building 15 room 253 (zoom link)

Host: Antoine Islegen-Wojdyla (Advanced Light Source/Berkeley Lab)

Description: The Advanced Photon Source Upgrade (APS-U) represents a transformative leap in synchrotron radiation capabilities, delivering unprecedented X-ray brightness and coherence. These advances open new frontiers in materials science, biology, and nanotechnology by enabling high-resolution imaging and real-time analysis of complex systems. Fully harnessing these capabilities, however, requires the X-ray beam to be maintained with exceptional precision and stability. To address this challenge, the Optics Group at the APS developed and deployed AutoFocus, an AI-driven software suite that autonomously aligns and optimizes X-ray nanofocusing optics with remarkable speed and accuracy while minimizing wavefront distortions and preserving beam coherence. AutoFocus can integrate with any radiation-detection and beam-characterization technique available at the newly commissioned APS-U beamlines, including our real-time, single-shot, reference-free wavefront sensor based on a coded mask. This integration enables autonomous optical alignment guided directly by the measured wavefront properties. A crucial component of these developments is the implementation of accurate and realistic digital twins of the optical systems using the simulation libraries of the widely adopted OASYS suite. These digital twins provide a safe environment in which to develop and test the software without risking beamline hardware or operations. They can also serve as a powerful source of low-fidelity datasets for preconfiguring and pretraining the AI, thereby reducing the amount of valuable beamtime required for these activities.