ERR@HRI 2.0 Challenge: Multimodal Detection of Errors and Failures in Human-Robot Conversations

TL;DR

The paper introduces ERR@HRI 2.0, a multimodal benchmark for detecting errors in LLM-powered human–robot conversations. It defines two perspectives on robot errors (system and user) and provides a dataset with facial, head pose, and audio features plus speech embeddings from two embodied agents across five tasks. The study establishes a comprehensive evaluation framework with offline and streaming metrics, and provides baseline models (notably a random forest) that establish initial performance benchmarks. The work aims to advance multimodal failure detection in HRI, enabling more robust and trustworthy human–robot interactions, and invites broader participation and future extensions to include more modalities and environments.

Abstract

The integration of large language models (LLMs) into conversational robots has made human-robot conversations more dynamic. Yet, LLM-powered conversational robots remain prone to errors, e.g., misunderstanding user intent, prematurely interrupting users, or failing to respond altogether. Detecting and addressing these failures is critical for preventing conversational breakdowns, avoiding task disruptions, and sustaining user trust. To tackle this problem, the ERR@HRI 2.0 Challenge provides a multimodal dataset of LLM-powered conversational robot failures during human-robot conversations and encourages researchers to benchmark machine learning models designed to detect robot failures. The dataset includes 16 hours of dyadic human-robot interactions, incorporating facial, speech, and head movement features. Each interaction is annotated with the presence or absence of robot errors from the system perspective, and perceived user intention to correct for a mismatch between robot behavior and user expectation. Participants are invited to form teams and develop machine learning models that detect these failures using multimodal data. Submissions will be evaluated using various performance metrics, including detection accuracy and false positive rate. This challenge represents another key step toward improving failure detection in human-robot interaction through social signal analysis.

Figures (1)

• Figure 1: Illustration of the interaction and data recording setup, together with extracted features.