Explaining Autonomy: Enhancing Human-Robot Interaction through Explanation Generation with Large Language Models

TL;DR

The paper tackles explainability in human–robot interaction by proposing a retrieval-augmented, LLM-driven explanation system for autonomous robots. It formalizes a log-based explanation framework (XAR) that uses a vector database and an LLM to generate contextual explanations from ROS 2 logs, aided by a retrieval mechanism to supply relevant information. The approach is evaluated on a real-world ERL navigation benchmark, with both objective metrics (log processing, response times) and a user questionnaire to assess explanation quality. Findings suggest LLMs can produce useful explanations from logs, but performance hinges on log verbosity and effective retrieval, indicating avenues for refinement before broad deployment. The work advances practical explainability in robotics and provides a foundation for adapting explanations to different user expertise and tasks beyond navigation.

Abstract

This paper introduces a system designed to generate explanations for the actions performed by an autonomous robot in Human-Robot Interaction (HRI). Explainability in robotics, encapsulated within the concept of an eXplainable Autonomous Robot (XAR), is a growing research area. The work described in this paper aims to take advantage of the capabilities of Large Language Models (LLMs) in performing natural language processing tasks. This study focuses on the possibility of generating explanations using such models in combination with a Retrieval Augmented Generation (RAG) method to interpret data gathered from the logs of autonomous systems. In addition, this work also presents a formalization of the proposed explanation system. It has been evaluated through a navigation test from the European Robotics League (ERL), a Europe-wide social robotics competition. Regarding the obtained results, a validation questionnaire has been conducted to measure the quality of the explanations from the perspective of technical users. The results obtained during the experiment highlight the potential utility of LLMs in achieving explanatory capabilities in robots.

Figures (15)

• Figure 1: Explainability system workflow. Use of robot's logs and RAG method to generate explanations.
• Figure 2: Explainability framework components
• Figure 3: Implementation of an explainability system based on the proposed framework.
• Figure 4: Testbed map for the experiment
• Figure 5: Self-Identification Among Participants

Theorems & Definitions (1)

• Proposition 1