FASTER PROJECT

ERUM-DATA PROGRAM

Autonomous Experiments &
Fast Data Pipelines

Unlocking the potential of synchrotron radiation and neutron science through autonomous control systems, intelligent decision-making, and FPGA-based hardware acceleration.

FASTER Closed Loop Architecture
Closed-Loop Autonomous ExperimentationInspired by [Noack2019]
01

Autonomous Workflows

Self-adjusting parameters based on rapid, automated data analysis. Overcoming the limits of traditional manual control systems for complex phase diagram exploration.

02

Fast Feedback Loop

Real-time data pipelines for automatic background subtraction, noise reduction, and beam damage detection linked directly to end-station control systems.

03

Intelligent Optimization

Implementing Bayesian optimization and reinforcement learning to navigate multi-dimensional experimental landscapes and maximize data quality.

Hardware-Accelerated
Computational Pipelines

✓

FPGA Hardware Acceleration

Meeting real-time feedback demands where conventional CPU or ML algorithms are too slow. AI accelerators integrated directly into hardware.

✓

Multi-Tau Correlation

Running computationally intensive analyses directly on chip to enable low-latency inference for XPCS and other high-volume use cases.

✓

Unified Control Frameworks

Integration into established beamline environments like BLISS and Sardana to ensure modularity and transferability across research facilities.

P08
PETRA III, DESY

X-ray Photon Correlation Spectroscopy focus.

P10
PETRA III, DESY

X-ray and Neutron reflectivity exploration.

RS
REFSANS, MLZ

Time-resolved Reciprocal Space Mapping (TR-RSM).

+
Transferability

Designing solutions for global synchrotron applications.

Strategic Impact & Funding Goals

Our project aligns with the key funding policy goals for large-scale research infrastructures.

Accelerating materials discovery via high-throughput screening.
Optimizing beamtime efficiency at large-scale infrastructures.
Enhancing data quality through reduced bias and real-time noise reduction.
Building cross-disciplinary bridges between Physics, Electrical Engineering, and AI.
Enabling non-expert and industrial user access through standardized workflows.
Continuous refinement via curated Machine Learning datasets and the ErUM-Data Hub.

Interdisciplinary Consortium

Our consortium brings together world-class experts from synchrotron and neutron science, accelerator technology, electrical engineering, and industry leaders to pioneer a new era of autonomous experimentation.

UoS
DESY
MLZ
DAPHNE
DIG-UM