Drivers' mental models of advanced driver assistance systems: A systematic review of conceptualization, associated factors, and intervention strategies

초록

Drivers' mental models of advanced driver assistance systems (ADAS) are their internal representations of how ADAS operate, encompassing an understanding of system capabilities, limitations, and contextual constraints. Well-calibrated mental models of ADAS are essential for safe and effective use, especially in critical situations that approach the operational boundaries. Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, this systematic review synthesizes findings from 71 empirical studies to examine three primary aspects: the conceptualization and measurement of drivers' ADAS mental models; the factors associated with variations in drivers' ADAS mental models; and the interventions proposed to enhance drivers' system understanding. The review reveals considerable heterogeneity in both the terminology and methodologies employed across studies. While mental model accuracy consistently correlates with visual attention metrics, its relationship with vehicle control performance is more variable. Individual differences (e.g., demographics, subjective perceptions, prior ADAS exposure, and information sources) contribute to variation in mental models, although effects are sometimes inconsistent. Training strategies are broadly classified into three categories: expository, interactive, and hands-on. Each of these has demonstrated effectiveness under particular conditions, with integrated methods often proving more beneficial. Humanmachine interfaces that support the dynamic updating of mental models are also reviewed. Findings underscore the need for clearer distinctions between general and applied mental models to improve conceptual clarity and methodological comparability. They also highlight the importance of longitudinal research in evaluating the durability of training effects. A further need is identified for standardized frameworks in the design of experimental scenarios and outcome assessments.

키워드