A-DOT (Active Deployable Optical Telescope) is a payload prototype project of a 6U segmented deployable telescope with an aperture diameter of 300 mm currently in the design phase. This paper investigates two different strategies for phasing a deployable segmented telescope. The first method employs a classical optimisation approach, where the image sharpness is used as the primary metric for aligning the mirror segments. This technique involves iteratively adjusting the individual segments' positions and orientations to maximise the resulting image's sharpness. The second method takes a more innovative approach by leveraging the power of deep learning techniques. Deep learning algorithms, trained on a large dataset of simulated images, can learn to recognise and correct phasing errors automatically. This approach can potentially streamline the phasing process and enhance the telescope's overall performance. Preliminary results from the study demonstrate the efficacy of both methods in achieving excellent phasing control. Remarkably, these techniques have successfully identified and corrected significant phasing errors, with path length differences of several microns, ultimately reducing the residual errors to the desired performance level using a point source, typically below 15 nm in the visible spectrum.
|