Development and Evaluation of a Chatbot to Support Pre-Mission Planning in a Launch and Re-entry Coordination Center Using Retrieval-Augmented Generation (RAG)

The growing number of orbital launch and re-entry operations demands precise and well-coordinated planning, particularly by institutions such as a Launch and Re-entry Coordination Center. In the pre-mission phase, fast and reliable access to mission-critical knowledge is essential to ensure safe and efficient coordination. This work presents the development of an AI-based chatbot that supports planning activities by providing relevant information on demand. The chatbot utilizes Retrieval-Augmented Generation (RAG), a technique that combines generative language models with document-based information retrieval. This allows for the generation of context-specific responses grounded in domain-specific resources such as coordination procedures, mission documentation, and regulatory requirements. The technical implementation is based on a modular RAG stack consisting of a vector database, retriever, and Lage Language Model (LLM) component. The evaluation focuses on system architecture, response quality, context relevance, and user acceptance. Initial tests in a simulated coordination environment indicate that the chatbot can reliably answer typical pre-mission planning queries, contributing to more efficient workflows and error reduction. The results highlight the potential of LLM-based assistance systems in safety-critical space coordination tasks and provide a basis for further automation in mission planning and operational support.