Adaptive optics for compensating for wavefront aberrations have been reported in astronomy, optical coherence tomography, microscope imaging, and laser material processing because the aberrations cause focal spot distortion leading to loss of resolution and efficiency in these applications. In adaptive optics, wavefront sensing is important. Several approaches for wavefront sensing have been demonstrated, such as a Shack–Hartmann wavefront sensor, shearing interferometry, the Transport-of-Intensity Equation (TIE), and iterative algorithms for phase retrieval. In this presentation, femtosecond laser processing with aberration compensation based on machine learning was demonstrated. The aberrations existing in the laser processing were continuously predicted by the trained neural network with an update period of 36 ms and were compensated by a spatial light modulator. In the experiment, the neural network-based adaptive optics reduced the wavefront error in the laser processing to most one-ninth. Therefore, the adaptive optics improved the resolution in laser processing.
|