FASTER

Towards fast feedback mechanism and autonomous experiments in synchrotron and neutron science

01/11/2025 to 31/10/2028

Our project aims to develop autonomous and unattended experiments in synchrotron and neutron research
to accelerate scientific discoveries and make more efficient use of beam time. Systems that independently
adjust experimental parameters through fast data analysis overcome the limitations of manual control. This
enables more precise investigations of complex phase diagrams, effective high-throughput screenings, and
improved data quality by reducing radiation damage and systematic errors.
An initial focus is the implementation of autonomous workflows at the P08 and P10 beamlines at PETRA III
(DESY) and REFSANS at MLZ/FRM II, with applications such as XPCS, reflectivity, and pump-probe
experiments. These workflows are intended to be extended to neutron sources such as ESS and ILL. At the
core of the project lies an autonomous data pipeline for background subtraction, mask selection, beam
damage detection, and parameter estimation. Optimization algorithms and FPGA-based hardware
acceleration ensure real-time feedback, enabling dynamic experiment control closely integrated with the
beamline.
Additionally, a database will be created to provide machine learning-oriented training datasets. This
resource supports the continuous improvement of optimization routines and ensures system scalability. The
results will be made accessible through the ErUM-Data Hub to foster scientific collaboration. At the same
time, an industrial partnership will be pursued to make the developed technologies applicable in other fields.
Our project takes the first steps toward autonomous experiments in synchrotron and neutron research,
aiming to enhance research efficiency and data quality.