SAGE-LLM: Towards Safe and Generalizable LLM Controller with Fuzzy-CBF Verification and Graph-Structured Knowledge Retrieval for UAV Decision

2026-02-27

AI Summary

This work addresses the limitations of large language models (LLMs) in autonomous drone decision-making, particularly their lack of domain-specific knowledge and formal safety guarantees in complex, dynamic environments with unknown threats. To overcome these challenges, the authors propose SAGE-LLM, a two-tier LLM-based decision architecture that operates without online training and integrates high-level semantic planning with low-level precise control. The key innovations include a fuzzy control barrier function (Fuzzy-CBF) to formally verify the safety of LLM-generated actions and a star-shaped hierarchical graph-based retrieval-augmented generation (RAG) mechanism for efficient, interpretable context-aware knowledge retrieval. Evaluated on pursuit-evasion tasks involving unknown obstacles and sudden threats, SAGE-LLM significantly enhances safety and generalization while maintaining task performance, demonstrating strong potential for extension to other embodied intelligent systems.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Planning, Routing, and Scheduling: Planning with Language Models
• Natural Language Processing: Safety and Robustness

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

In UAV dynamic decision, complex and variable hazardous factors pose severe challenges to the generalization capability of algorithms. Despite offering semantic understanding and scene generalization, Large Language Models (LLM) lack domain-specific UAV control knowledge and formal safety assurances, restricting their direct applicability. To bridge this gap, this paper proposes a train-free two-layer decision architecture based on LLMs, integrating high-level safety planning with low-level precise control. The framework introduces three key contributions: 1) A fuzzy Control Barrier Function verification mechanism for semantically-augmented actions, providing provable safety certification for LLM outputs. 2) A star-hierarchical graph-based retrieval-augmented generation system, enabling efficient, elastic, and interpretable scene adaptation. 3) Systematic experimental validation in pursuit-evasion scenarios with unknown obstacles and emergent threats, demonstrating that our SAGE-LLM maintains performance while significantly enhancing safety and generalization without online training. The proposed framework demonstrates strong extensibility, suggesting its potential for generalization to broader embodied intelligence systems and safety-critical control domains.

Problem

• UAV decision
• Large Language Models
• safety assurance
• generalization
• hazardous factors

Innovation

• Fuzzy Control Barrier Function
• Graph-Structured Knowledge Retrieval
• Retrieval-Augmented Generation
• Safety Verification
• Embodied Intelligence