Advancing Shared and Multi-Agent Autonomy in Underwater Missions: Integrating Knowledge Graphs and Retrieval-Augmented Generation

2025-07-27

AI Summary

Underwater environments severely impede robotic autonomy due to low visibility, strong disturbances, and constrained communication. To address these challenges, this work proposes a large language model (LLM) architecture integrating knowledge graphs (KGs) and retrieval-augmented generation (RAG), establishing the first semantic-driven shared autonomy framework for underwater multi-agent systems. The approach injects domain-specific taxonomies and structured environmental priors into the LLM, enabling context-aware reasoning, cross-agent collaborative decision-making, and natural-language human–robot interaction. KG-based constraints mitigate hallucination, ensuring decision consistency and interpretability. In real-world underwater missions, the system achieves 100% task success rate and behavioral completeness. Ablation studies confirm that structured domain knowledge is critical for eliminating decision bias and enhancing robustness. This work bridges semantic reasoning with embodied autonomy in challenging aquatic settings, advancing trustworthy AI for underwater robotics.

Technology Category

• Intelligent Robots: Embodied AI
• Multiagent Systems: Multiagent Systems under Uncertainty
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Agentic search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Robotic platforms have become essential for marine operations by providing regular and continuous access to offshore assets, such as underwater infrastructure inspection, environmental monitoring, and resource exploration. However, the complex and dynamic nature of underwater environments, characterized by limited visibility, unpredictable currents, and communication constraints, presents significant challenges that demand advanced autonomy while ensuring operator trust and oversight. Central to addressing these challenges are knowledge representation and reasoning techniques, particularly knowledge graphs and retrieval-augmented generation (RAG) systems, that enable robots to efficiently structure, retrieve, and interpret complex environmental data. These capabilities empower robotic agents to reason, adapt, and respond effectively to changing conditions. The primary goal of this work is to demonstrate both multi-agent autonomy and shared autonomy, where multiple robotic agents operate independently while remaining connected to a human supervisor. We show how a RAG-powered large language model, augmented with knowledge graph data and domain taxonomy, enables autonomous multi-agent decision-making and facilitates seamless human-robot interaction, resulting in 100% mission validation and behavior completeness. Finally, ablation studies reveal that without structured knowledge from the graph and/or taxonomy, the LLM is prone to hallucinations, which can compromise decision quality.

Problem

• Enhancing autonomy in underwater missions with multi-agent systems
• Improving decision-making using knowledge graphs and RAG systems
• Ensuring reliable human-robot interaction in dynamic underwater environments

Innovation

• Knowledge graphs enhance underwater data structuring
• Retrieval-augmented generation improves decision-making
• Multi-agent autonomy with human oversight