Enhancing Autonomous Ship Communication: A Cost-Effective and High-Accuracy LLM Framework Using Decision Trees and RAG

This study introduces a novel architecture designed to enhance the performance and cost-efficiency of Large Language Models (LLMs) in autonomous ship communication systems. Autonomous ships require high accuracy and rapid response, yet their operational constraints-limited computational resources and lack of internet connectivity-pose significant challenges for traditional LLMs. To address these issues, we integrate Retrieval-Augmented Generation (RAG) with Decision Trees. This integration improves the efficiency of RAG's data retrieval and processing, significantly reduces the computational overhead of LLMs, and enhances response accuracy. Evaluations using 200 hours of real-world maritime communication data demonstrate that the proposed system outperforms existing methods in speed and accuracy under resource-constrained conditions. This research advances the practical application and reliability of LLMs in autonomous ship communication, providing a strong foundation for improving automation and ensuring safety in the maritime industry.