A critical review on the state-of-the-art and future prospects of Machine Learning for Earth Observation Operations

The continuing Machine Learning (ML) revolution indubitably has had a significant positive impact on the analysis of downlinked satellite data. Other aspects of the Earth Observation industry, despite being less susceptible to widespread application of Machine Learning, are also following this trend. These applications, actual use cases, possible prospects and difficulties, as well as anticipated research gaps, are the focus of this review of Machine Learning applied to Earth Observation Operations. A wide range of topics are covered, including mission planning, fault diagnosis, fault prognosis and fault repair, optimization of telecommunications, enhanced GNC, on-board image processing, and the use of Machine Learning models on platforms with constrained compute and power capabilities, as well as recommendations in the respective areas of research. The review tackles all on-board and off-board applications of machine learning to Earth Observation with one notable exception: it omits all post-processing of payload data on the ground, a topic that has been studied extensively by past authors. In addition, this review article discusses the standardization of Machine Learning (i.e., Guidelines and Roadmaps), as well as the challenges and recommendations in Earth Observation operations for the purpose of building better space missions.