Trust Modeling and Estimation in Human-Autonomy Interactions

TL;DR

This work addresses the challenge of modeling human supervisor trust in human-autonomy interactions with asymmetrical and intermittently communicated trust dynamics. It introduces a switched-linear trust model with memory and state saturation, coupled with event-triggered sampling, and demonstrates parameter identification via data from a 51-participant user study. The study further shows that clustering individuals into ambivalent, pessimistic, and optimistic trust groups improves predictive accuracy over population or individual models, with first-order models often adequate. The approach enables real-time trust estimation and has potential to inform onboard observers that steer autonomous system behavior to maintain effective collaboration and reduce supervision load. The findings highlight distinct trust profiles and provide a principled framework for adaptive human-machine interfaces.

Abstract

Advances in the control of autonomous systems have accompanied an expansion in the potential applications for autonomous robotic systems. The success of applications involving humans depends on the quality of interaction between the autonomous system and the human supervisor, which is particularly affected by the degree of trust that the supervisor places in the autonomous system. Absent from the literature are models of supervisor trust dynamics that can accommodate asymmetric responses to autonomous system performance and the intermittent nature of supervisor-autonomous system communication. This paper focuses on formulating an estimated model of supervisor trust that incorporates both of these features by employing a switched linear system structure with event-triggered sampling of the model input and output. Trust response data collected in a user study with 51 participants were then used identify parameters for a switched linear model-based observer of supervisor trust.

Figures (13)

• Figure 1: Overview of the trust framework.
• Figure 2: The proposed six modes of trust.
• Figure 3: The coefficients of autonomous system performance for $\sigma\in\{1,2,3\}$; note that the graph is identical for $\sigma\in\{4,5,6\}$ with $B_5$ substituted for $B_2$.
• Figure 4: The user study's simulation interface.
• Figure 5: Autonomous system performance and predicted supervisor trust for participant 23; — ground truth, + first-order model ($n_q^*=120.0$ s), x second-order model ($n_q^*=120.0$ s).