Language Modeling and Retrieval-Augmented Generation for Integration in Space Operation Decision Support Tools
Published in 58th IAA Symposium on Safety, Quality and Knowledge Management in Space Activities, 2025
Recommended citation: Ruben Belo, Claudia Soares, Marta Guimaraes, "Language Modeling and Retrieval-Augmented Generation for Integration in Space Operation Decision Support Tools." 58th IAA Symposium on Safety, Quality and Knowledge Management in Space Activities, 2025. http://dx.doi.org/10.52202/083095-0016
Abstract
The rapid expansion of the space industry has led to an increasing volume of technical documents, operational guidelines, and scientific publications. As space activities become more complex, effective decision-making and operations management require efficient processing of vast and ever-changing information sources to support new developments in the sector. Space operators must navigate regulations, guidelines, safety protocols, and mission-critical data, which makes knowledge retrieval a significant challenge. The ability to extract, synthesise, and communicate relevant information quickly is essential for ensuring safety, optimizing operations, and addressing emerging challenges in space traffic management and debris mitigation. To address these challenges, we propose a conversational agent powered by a Retrieval Augmented Generation (RAG) pipeline, leveraging Large Language Models (LLMs) and information retrieval techniques. Our system processes extensive documentation, retrieving and generating contextually relevant responses based on official reports, scientific literature, operational directives, and guidelines to provide actionable insights for space operations. Our methodology involves evaluating various retrieval strategies and embedding techniques to improve information accuracy. Furthermore, we explore different RAG configurations to optimize response generation, ensuring precise and reliable communication. To mitigate the risk of generating misinformation and harmful content, we integrate state-of-the-art techniques for conditioning LLM outputs, reinforcing safety and ethical considerations in our conversational agent’s responses. This research integrates state-of-the-art AI technologies with domain-specific space knowledge to contribute to the development of intelligent systems for space operations. Our approach aims to improve decision-making and space traffic management, and address the growing challenges posed by space debris. Ultimately, this research supports the advancement of safer and more sustainable space activities and applications, aligning with global efforts to ensure long-term space exploration.
BibTeX
@inproceedings{Belo_2025,
author = "Belo, Ruben and Soares, Claudia and Guimaraes, Marta",
series = "IAC 2025",
title = "Language Modeling and Retrieval-Augmented Generation for Integration in Space Operation Decision Support Tools",
url = "http://dx.doi.org/10.52202/083095-0016",
DOI = "10.52202/083095-0016",
booktitle = "58th IAA Symposium on Safety, Quality and Knowledge Management in Space Activities",
publisher = "International Astronautical Federation (IAF)",
year = "2025",
pages = "144-155",
collection = "IAC 2025",
abstract = "The rapid expansion of the space industry has led to an increasing volume of technical documents, operational guidelines, and scientific publications. As space activities become more complex, effective decision-making and operations management require efficient processing of vast and ever-changing information sources to support new developments in the sector. Space operators must navigate regulations, guidelines, safety protocols, and mission-critical data, which makes knowledge retrieval a significant challenge. The ability to extract, synthesise, and communicate relevant information quickly is essential for ensuring safety, optimizing operations, and addressing emerging challenges in space traffic management and debris mitigation. To address these challenges, we propose a conversational agent powered by a Retrieval Augmented Generation (RAG) pipeline, leveraging Large Language Models (LLMs) and information retrieval techniques. Our system processes extensive documentation, retrieving and generating contextually relevant responses based on official reports, scientific literature, operational directives, and guidelines to provide actionable insights for space operations. Our methodology involves evaluating various retrieval strategies and embedding techniques to improve information accuracy. Furthermore, we explore different RAG configurations to optimize response generation, ensuring precise and reliable communication. To mitigate the risk of generating misinformation and harmful content, we integrate state-of-the-art techniques for conditioning LLM outputs, reinforcing safety and ethical considerations in our conversational agent’s responses. This research integrates state-of-the-art AI technologies with domain-specific space knowledge to contribute to the development of intelligent systems for space operations. Our approach aims to improve decision-making and space traffic management, and address the growing challenges posed by space debris. Ultimately, this research supports the advancement of safer and more sustainable space activities and applications, aligning with global efforts to ensure long-term space exploration.",
abstractsource = "https://blog.neuraspace.com/hubfs/Language-Modeling-and-Retrieval-Augmented-Generation-Integrationin-Space-Operation\%20Decision-Support-Tools.pdf",
abstractsourcelabel = "Published abstract source"
}