Dynamic Interferometry for Freeform Surface Measurement Based on Machine Learning-Configured Deformable Mirror.

Abstract

Optical freeform surfaces are widely used in imaging and non-imaging systems due to their high design freedom. In freeform surface manufacturing and assembly, dynamic freeform surface measurement that can guide the next operation remains a challenge. To meet this urgent need, we propose a dynamic interferometric method based on a machine learning-configured deformable mirror (DM). In this method, a dynamic interferometric system is developed. By using coaxial structure and polarization interference, transient measurement of the measured surface can be realized to meet dynamic requirements, and at the same time, DM transient monitoring can be realized to reduce the accuracy loss caused by DM surface changes and meet dynamic requirements. A transient phase modulation scheme using machine learning to configure the DM surface is proposed, which keeps the system in a measurable state. Compared with the traditional phase modulation scheme that relies on iteration, the scheme proposed in this paper is more efficient and is conducive to meeting dynamic requirements. The feasibility is verified by practical experiments. The research in this paper has significance for guiding the application of dynamic interferometry in the measurement of dynamic surfaces.