AI-Corrected Vision for the James Webb Space Telescope
Researchers at the University of Sydney have developed an AI-powered calibration system that restored the James Webb Space Telescope’s interferometric instrument (AMI) without physical intervention.
Their software framework, AMIGO (Aperture Masking Interferometry Generative Observations), employs neural networks and simulation-based modeling to mitigate the brighter-fatter effect, a charge migration phenomenon that caused subtle blurring in JWST images.
The algorithm conducts data-driven deconvolution and point-spread function modeling to reconstruct high-fidelity images, restoring sub-pixel precision and enabling clearer detections of exoplanets, stellar winds, and galactic structures.
This represents a major advance in software-defined optics, demonstrating that mission-critical image correction can be achieved entirely via AI.
Primary Research:
- Desdoigts, L. et al. “AMIGO: a data-driven calibration of the JWST interferometer,” arXiv:2510.09806 (2025). Read on arXiv →
- Charles, M. et al. “Image reconstruction with the JWST interferometer,” arXiv:2510.10924 (2025). Read on arXiv →