A Collaborative Reasoning Framework for Anomaly Diagnostics in Underwater Robotics

2025-11-04

AI Summary

Underwater robots struggle to handle unforeseen anomalies in safety-critical scenarios. Method: This paper proposes a human-in-the-loop real-time anomaly diagnosis framework integrating large language models (LLMs), high-fidelity digital twins, and a dual-agent architecture: one agent performs multi-source signal-driven anomaly detection, while the other conducts natural-language-based root-cause reasoning; interactive dynamic knowledge distillation continuously injects expert knowledge into the system. Contribution/Results: The framework establishes a closed-loop “perception–diagnosis–feedback” pipeline, enabling structured signal interpretation and knowledge-augmented dialogue. Experiments demonstrate significant improvements in anomaly detection accuracy, interpretability of root-cause attribution, and self-evolution capability of low-level perception models. This advances AI from a static tool toward an adaptive, collaborative partner for underwater robotic systems.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Humans and AI: Human-in-the-loop Machine Learning
• Natural Language Processing: Safety and Robustness

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• 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

Abstract

The safe deployment of autonomous systems in safety-critical settings requires a paradigm that combines human expertise with AI-driven analysis, especially when anomalies are unforeseen. We introduce AURA (Autonomous Resilience Agent), a collaborative framework for anomaly and fault diagnostics in robotics. AURA integrates large language models (LLMs), a high-fidelity digital twin (DT), and human-in-the-loop interaction to detect and respond to anomalous behavior in real time. The architecture uses two agents with clear roles: (i) a low-level State Anomaly Characterization Agent that monitors telemetry and converts signals into a structured natural-language problem description, and (ii) a high-level Diagnostic Reasoning Agent that conducts a knowledge-grounded dialogue with an operator to identify root causes, drawing on external sources. Human-validated diagnoses are then converted into new training examples that refine the low-level perceptual model. This feedback loop progressively distills expert knowledge into the AI, transforming it from a static tool into an adaptive partner. We describe the framework's operating principles and provide a concrete implementation, establishing a pattern for trustworthy, continually improving human-robot teams.

Problem

• Develops collaborative framework for anomaly diagnostics in robotics
• Integrates AI models with human expertise for real-time response
• Creates adaptive system that converts expert knowledge into training

Innovation

• Integrates LLMs with digital twin for diagnostics
• Uses dual agents for anomaly monitoring and reasoning
• Implements human-validated feedback loop for adaptation