Advancing Earth Observation: A Survey on AI-Powered Image Processing in Satellites
Aidan Duggan, Bruno Andrade, Haithem Afli
TL;DR
The paper addresses the challenge of processing Earth Observation imagery onboard satellites amid strict power, compute, memory, and radiation constraints. It surveys current literature, outlining the key constraints and practical mitigation strategies, including hardware accelerators, model optimization, and onboard data management. It synthesizes real-world demonstrations (e.g., CloudScout on Eyes of Things, Jetson-based on-board processing) and traces the evolution of space-grade computing architectures. The findings highlight the potential to dramatically reduce downlink needs and enable autonomous, real-time EO analytics, while calling for more quantitative assessments and broader deployments.
Abstract
Advancements in technology and reduction in it's cost have led to a substantial growth in the quality & quantity of imagery captured by Earth Observation (EO) satellites. This has presented a challenge to the efficacy of the traditional workflow of transmitting this imagery to Earth for processing. An approach to addressing this issue is to use pre-trained artificial intelligence models to process images on-board the satellite, but this is difficult given the constraints within a satellite's environment. This paper provides an up-to-date and thorough review of research related to image processing on-board Earth observation satellites. The significant constraints are detailed along with the latest strategies to mitigate them.