Laser Focal Position Correction Using FPGA-based ML models

High repetition-rate, ultrafast laser systems are essential to modern scientific and industrial applications. Variations in critical figures of merit, such as focal position, can significantly impact efficacy for applications involving laser plasma interactions, such as electron beam acceleration and radiation generation. We present a diagnostic and correction scheme for controlling and determining laser focal position by utilizing fast wavefront sensor measurements from multiple positions to train a focal position predictor. This predictor and additional control algorithms are integrated into a unified control interface and FPGA-based controller for high power operation at operating beamlines at the BELLA Center at Lawrence Berkeley National Laboratory. Online optical adjustments are made to a telescopic lens to provide the desired correction to the focal position on millisecond timescales. Our initial proof-of-principle demonstrations leverage pre-compiled data and pre-trained networks operating ex-situ from the laser system. A framework for generating a low-level hardware description of ML-based correction algorithms on FPGA hardware is coupled directly to the beamline using the AMD Xilinx Vitis AI toolchain in conjunction with deployment scripts. Lastly, we consider the use of remote computing resources, such as the Sirepo scientific framework, to actively update these correction schemes and deploy models to a production environment.