Quantitative Trust and Behavioral Measurement Tools for Optimizing Human–Robot Interaction

Quantifying trust in human–robot interaction is a significant challenge due to the lack of a universally accepted definition of “trust” in this domain. Existing approaches often rely on static self-report measures, which are limited in capturing dynamic trust behaviors in safety critical contexts such as advanced air mobility. To address this gap, we introduce a novel empirical framework for quantifying trust in this domain. The proposed framework integrates performance, behavioral, and data-driven metrics to assess how users calibrate their trust during simulated flight missions. Four distinct trust-measurement tools are presented: efficiency factor, obstacle clearance factor, manual/automatic mode usage, and geometric difference. In a user study with 40 participants across 30 simulation runs, the results show that 1) trust in the automation increases with operator experience, as evidenced by improved efficiency ($r = -0.23$,$p < 10^{-12}$) and fewer manual interventions ($r = -0.22$,$p = 7.6\times 10^{-15}$); 2) stable individual differences (e.g., risk tolerance) significantly influence trust-related metrics, with obstacle clearance distances remaining consistent for most participants across runs ($r = -0.13$,$p = 1.9\times 10^{-5}$); and 3) the geometric difference metric captures nuanced behavioral variations, revealing distinct trust-based operational strategies among users. By integrating performance metrics and behavioral patterns, this multidimensional framework offers practical tools for enhancing the trustworthiness of autonomous systems. It fills a key gap in current trust assessment approaches, supports the development of safer and more effective advanced air mobility systems, and contributes to the broader advancement of human–robot interaction research.