Modeling and Evaluating Trust Dynamics in Multi-Human Multi-Robot Task Allocation

2024-09-24

AI Summary

In human–human–multi-robot (MH-MR) collaborative task allocation, neglecting dynamic trust evolution leads to suboptimal efficiency and robustness. Method: This paper proposes the Expectation Confirmation Trust (ECT) model—the first to systematically formalize cross-entity (human–human, human–robot, robot–robot), time-varying, and bidirectional trust dynamics—grounded in expectation confirmation theory and multi-agent modeling. We develop a scalable, dynamic trust evaluation framework and integrate it into a simulation platform featuring five classical trust models plus a trust-agnostic baseline for rigorous comparative evaluation. Contribution/Results: Across diverse configurations (2H–2R to 10H–10R), ECT-driven task allocation significantly improves task success rate, reduces average completion time, and lowers error rates. These results empirically validate that explicit, dynamic trust modeling yields substantial gains in system robustness and collaborative performance.

Technology Category

• Intelligent Robots: Multi-Robot Systems
• Humans and AI: Planning and Decision Support for Human-Machine Teams
• Multiagent Systems: Multiagent Systems under Uncertainty

Application Category

• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Metrics for user behavior and evaluating success

Abstract

Trust is essential in human-robot collaboration, particularly in multi-human, multi-robot (MH-MR) teams, where it plays a crucial role in maintaining team cohesion in complex operational environments. Despite its importance, trust is rarely incorporated into task allocation and reallocation algorithms for MH-MR collaboration. While prior research in single-human, single-robot interactions has shown that integrating trust significantly enhances both performance outcomes and user experience, its role in MH-MR task allocation remains underexplored. In this paper, we introduce the Expectation Confirmation Trust (ECT) Model, a novel framework for modeling trust dynamics in MH-MR teams. We evaluate the ECT model against five existing trust models and a no-trust baseline to assess its impact on task allocation outcomes across different team configurations (2H-2R, 5H-5R, and 10H-10R). Our results show that the ECT model improves task success rate, reduces mean completion time, and lowers task error rates. These findings highlight the complexities of trust-based task allocation in MH-MR teams. We discuss the implications of incorporating trust into task allocation algorithms and propose future research directions for adaptive trust mechanisms that balance efficiency and performance in dynamic, multi-agent environments.

Problem

• Modeling trust dynamics in multi-human multi-robot teams.
• Evaluating trust models for improved task allocation outcomes.
• Incorporating trust to enhance team performance and efficiency.

Innovation

• Introduces Expectation Confirmation Trust (ECT) Model
• Evaluates ECT model against five trust models
• Improves task success rate and reduces errors