🤖 AI Summary
To address insufficient robot trustworthiness in human-robot coexistence scenarios, this paper proposes a novel accountable and explainable architecture integrating blockchain and large language models (LLMs). Specifically, it couples an Ethereum light client—ensuring immutable, tamper-proof logging—with a fine-tuned Llama-2 model to establish end-to-end traceability from low-level ROS 2 action auditing to high-level natural-language explanations. The method dynamically generates semantic explanations evaluated across coherence, accuracy, and understandability. Evaluated on three representative navigation tasks, it achieves 100% accountability data integrity, explanation accuracy ≥92%, and an average 37% increase in trust among non-expert users. This work presents the first systematic, blockchain-augmented LLM interpretability framework, delivering a verifiable, human-understandable, and auditable technical pathway for safety-critical human-robot interaction.
📝 Abstract
The deployment of autonomous agents in environments involving human interaction has increasingly raised security concerns. Consequently, understanding the circumstances behind an event becomes critical, requiring the development of capabilities to justify their behaviors to non-expert users. Such explanations are essential in enhancing trustworthiness and safety, acting as a preventive measure against failures, errors, and misunderstandings. Additionally, they contribute to improving communication, bridging the gap between the agent and the user, thereby improving the effectiveness of their interactions. This work presents an accountability and explainability architecture implemented for ROS-based mobile robots. The proposed solution consists of two main components. Firstly, a black box-like element to provide accountability, featuring anti-tampering properties achieved through blockchain technology. Secondly, a component in charge of generating natural language explanations by harnessing the capabilities of Large Language Models (LLMs) over the data contained within the previously mentioned black box. The study evaluates the performance of our solution in three different scenarios, each involving autonomous agent navigation functionalities. This evaluation includes a thorough examination of accountability and explainability metrics, demonstrating the effectiveness of our approach in using accountable data from robot actions to obtain coherent, accurate and understandable explanations, even when facing challenges inherent in the use of autonomous agents in real-world scenarios.