Leveraging Machine Learning in Spacecraft Construction: A Literature Review

Machine learning, a rapidly adapting and evolving field of artificial intelligence has sparked an interest in astrophysicists. Subsequently, in today’s world, despite the severe intricacies of constructing a spacecraft and operating one, machine learning has been implemented in both the construction and piloting of a spacecraft. This literature review will discuss and evaluate the implications of various prevalent ML applications in spacecraft design and propagation. Specifically, this paper will focus on trajectory planning, autonomous control systems, thermal modeling, predictive maintenance, and generative design. After having considered the pros and cons of each application, this paper concludes that the application of ML – with careful model training – can improve mission efficiency, reliability, and development speed. Conversely, this paper also acknowledges grave constraints such as limited data availability for model training, computational constraints, and ethical/sustainable issues. Overall, ML proposes a lot of hopeful scope for the future of aerospace engineering. If the presented limitations can be handled or controlled in the future, ML proffers more efficient spacecraft design and propulsion.