中国科学院南京天文光学技术研究所机构知识库

Embodied Intelligence-based optical astronomical telescopes represent a AI-driven approach to au tonomous perception, decision-making, and optimization of observational performance under dynamic environmental  conditions, will provide intelligent operation to the widely distributed unmanned astronomical telescopes. The percep tion of astronomical telescope observation performance, primarily manifested in the optical performance perception,  can be achieved through multi-field curvature wavefront sensors, wavefront slope sensors, and image quality sen sors. These devices measure physical quantities represented as point spread function (PSF) or wavefront aberration  data. By utilizing the perception data of optical performance, combined with end-to-end imaging simulation and dig ital twin technology, telescopes can achieve active optical correction and active alignment, intelligently maintaining  and optimizing optical imaging quality. With the further integration of multi-modal sensing, machine learning, and  large language model, embodied intelligence will drive the transition of telescopes from ”precision instruments” to  ”intelligent instruments,” providing new technological pathways for unmanned astronomical observatories.