GPU-based Adaptive Optics Real-time Control System

The adaptive optics system measures wavefront information in real time and uses conjugate compensation to achieve wavefront distortion compensation.Adaptive optics has high requirements on the operating frequency in many applications,especially in astronomical applications where operating frequencies above kHz are required.An efficient wavefront controller must be designed to achieve adaptive optics in the closed-loop feedback process.In this paper,we use GPU efficient parallel computing capability to improve the efficiency of adaptive optics control system.We designed a self-adaptive optical wavefront control method based on GPU and carried out the following research work.This paper combines the existing wavefront sensor parameters,designs the calculation process and control calculation process of adaptive optical wavefront measurement based on commercial wavefront sensors,analyzes the time complexity of wavefront slope calculation and reconstruction calculation and uses GPU to solve the problem.Adapt to the feasibility of optical control calculation process.Based on the analysis of computational complexity,the CUDA-based wavefront slope calculation process is designed.CUDA kernel is used to threshold the image and the computational task is decomposed,which ensures the calculation accuracy and improves the slope calculation efficiency.In the calculation process of wavefront reconstruction,the Zernike mapping matrix is solved in the data preprocessing stage,and the CUDA accelerating matrix operation improves the efficiency of wavefront reconstruction.At the same time,CUDA is used to process matrix operations in the wavefront control process to improve the wavefront correction efficiency of the system.This article analyzes and studies the process of adaptive optical control based on GPU.The developed piezoelectric deformable mirror is used as the wavefront corrector,and an adaptive optical system is built to perform adaptive optical closed-loop correction experiments.The system performance before and after using GPU optimization is tested and compared.Experimental results verify the feasibility of GPU in processing adaptive optics control system.Through experiments,the control calculation delay of the entire system is controlled within 2ms,which improves the system's working efficiency.At the same time,simulation experiments were performed on the 1000-element adaptive optics system to verify the feasibility of the GPU in processing multi-cell adaptive optics calculations and to provide reference for the development of control methods for large-aperture multi-cell adaptive optics systems.