Research Analyzer
← Back ICRA 2026

Estimating Trust in Human-Robot Collaboration through Behavioral Indicators and Explainability

Giulio Campagna, Marta Lagomarsino, Marta Lorenzini, Dimitrios Chrysostomou, Matthias Rehm, Arash Ajoudani

PDF

AI summary

Key figure (auto-extracted from paper)
Data-driven machine learning models accurately predict human trust preferences in real-time by analyzing dynamic behavioral indicators.
human-robot collaboration trust estimation behavioral indicators explainable AI machine learning Industry 5.0

Problem

Existing trust assessment methods rely on delayed surveys or single-factor models that fail to capture real-time, personalized trust dynamics during human-robot collaboration.

Approach

The framework extracts human motion and robot trajectory metrics during tasks, uses explicit user preferences to train machine learning classifiers, and applies SHAP explainability to identify key trust drivers.

Key results

  • Voting Classifier achieved 84.07% accuracy and 0.90 AUC-ROC for trust prediction
  • Framework successfully integrates human and robot behavioral indicators to predict trust preferences
  • SHAP analysis reveals individual variations in how behavioral cues influence trust
  • Optimized interaction parameters effectively enhance trust in industrial scenarios

Why it matters

Provides a practical, data-driven pathway for engineers to develop adaptive, human-centric robotic systems that dynamically adjust to operator trust levels for safer industrial collaboration.

Abstract

Industry 5.0 focuses on human-centric collaboration between humans and robots, prioritizing safety, comfort, and trust. This study introduces a data-driven framework to assess trust using behavioral indicators. The framework employs a Preference-Based Optimization algorithm to generate trust-enhancing trajectories based on operator feedback. This feedback serves as ground truth for training machine learning models to predict trust levels from behavioral indicators. The framework was tested in a chemical industry scenario where a robot assisted a human operator in mixing chemicals. Machine learning models classified trust with over 80% accuracy, with the Voting Classifier achieving 84.07% accuracy and an AUC-ROC score of 0.90. These findings under- score the effectiveness of data-driven methods in assessing trust within human-robot collaboration, emphasizing the valuable role behavioral indicators play in predicting the dynamics of human trust.

Index terms

Human Factors and Human-in-the-Loop Acceptability and Trust Human-Robot Collaboration

Related papers