Deep-learning-assisted, two-stage phase control method for high-power mode-programmable orbital angular momentum beam generation

Abstract

High-power mode-programmable orbital angular momentum (OAM) beams have received substantial attention in recent years. They are widely used in optical communication, nonlinear frequency conversion, and laser processing. To overcome the power limitation of a single beam, coherent beam combining (CBC) of laser arrays is used. However, in specific CBC systems used to generate structured light with a complex wavefront, eliminating phase noise and realizing flexible phase modulation proved to be difficult challenges. In this paper, we propose and demonstrate a two-stage phase control method that can generate OAM beams with different topological charges from a CBC system. During the phase control process, the phase errors are preliminarily compensated by a deep-learning (DL) network, and further eliminated by an optimization algorithm. Moreover, by modulating the expected relative phase vector and cost function, all-electronic flexible programmable switching of the OAM mode is realized. Results indicate that the proposed method combines the characteristics of DL for undesired convergent phase avoidance and the advantages of the optimization algorithm for accuracy improvement, thereby ensuring the high mode purity of the generated OAM beams. This work could provide a valuable reference for future implementation of high-power, fast switchable structured light generation and manipulation.

Deep Learning Piston Aberration Control of Fiber Laser Phased Array By Spiral Phase Modulation

“…In fact, the shape of modulation phase can be any asymmetric shape. However, by modulating the expected piston aberration of the sub-beams with a shape of spiral phase, the orbital angular momentum beam can be easily obtained, which has unique applications in many fields [15]…”

“…Researchers have demonstrated that the mapping from far-field image to near-field piston distribution can be directly established by CNN [6][7][8][9][10][11][12][13]. In 2019, Tianyue Hou et al used a single-frame defocused plane image as a data set and the Mean Square Error (MSE) as the evaluation function to coarse predict the piston aberration [8,15]. Then, they used the SPGD algorithm for further iteration to achieve a co-phase output effect of FLPA, according to CNN's output.…”

Coherent beam combining as a multidisciplinary education and training environment

“…With the rapid development of machine learning and wide application, it is straightforward to rise the proposal that whether machine learning can be employed for the phase control. By encouraging the graduate students to study the related knowledge and test the application potential in phase control, it is found that phase control can be achieved based on convolutional neural network (CNN) can work effectively, which provide a new solution that can generate the phase control signal rapidly without loss of precision [7,8]. Graduate students can also be well trained to master the fundamentals of machine learning.…”

Single step phase optimisation for coherent beam combination using deep learning

“…This neural network solution was then improved further by stochastic gradient descent using a camera at the focal plane. Hou et al 31 later showed that this approach can also be applied for the creation of orbital angular momentum beams, when using a ring of fibres. There have also been several key results in the field of deep reinforcement learning applied to coherent beam combination 32 , 33 , which is a technique that allows a neural network to learn by trial and error on the experimental setup or in a virtual environment 34 .…”