Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

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

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

TL;DR

The paper tackles online trust estimation in human–robot collaboration by integrating human and robot behavioral indicators with a Preference‑Based Optimization loop to collect trust‑related preferences. A machine learning ensemble, augmented with SHAP explainability, predicts trust preferences from indicator changes and supports adaptive robot trajectory parameterization in a chemical task. The Voting Classifier achieves 84.07% accuracy and an AUC of 0.90, with SHAP identifying Reaction Time and Task Attention as key drivers and revealing individualized indicator importance. The method enables online, personalized trust-aware adaptation of robot behavior, though validation across more tasks and deployment in real‑world settings remain as future directions.

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 underscore 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.

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

TL;DR

The paper tackles online trust estimation in human–robot collaboration by integrating human and robot behavioral indicators with a Preference‑Based Optimization loop to collect trust‑related preferences. A machine learning ensemble, augmented with SHAP explainability, predicts trust preferences from indicator changes and supports adaptive robot trajectory parameterization in a chemical task. The Voting Classifier achieves 84.07% accuracy and an AUC of 0.90, with SHAP identifying Reaction Time and Task Attention as key drivers and revealing individualized indicator importance. The method enables online, personalized trust-aware adaptation of robot behavior, though validation across more tasks and deployment in real‑world settings remain as future directions.

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 underscore 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.
Paper Structure (18 sections, 12 equations, 5 figures, 2 tables)

This paper contains 18 sections, 12 equations, 5 figures, 2 tables.

Table of Contents

  1. INTRODUCTION
  2. METHODOLOGY
  3. Definition of the Trust Indicators
  4. Human-Related Trust Indicators
  5. Robot-Related Trust Indicators
  6. Dataset Collection
  7. Preference Labels
  8. Data Pre-Processing and Trust Indicators Selection
  9. Machine Learning Classification of Human Preferences
  10. Model Explainability to Analyze the Influence of Behavioral Trust Indicators
  11. EXPERIMENTS
  12. Task Description and PBO Parameters
  13. Experimental Setup
  14. Experimental Protocol
  15. EXPERIMENTAL RESULTS
  16. ...and 3 more sections

Figures (5)

  • Figure 1: Illustration of HRC in an industrial setting, highlighting operator whole-body tracking (head and upper body) and interaction parameters guiding the robotic manipulator's behavior.
  • Figure 2: The experimental scenario with the robot assisting the human operator in mixing chemicals.
  • Figure 3: Evaluation of the Voting Classifier.
  • Figure 4: SHAP plot showing feature contributions to model predictions.
  • Figure 5: Importance of trust-related behavioral indicators for two participants, with X representing the mean SHAP values.