Towards a Safety-Centric Evaluation Framework for Augmented Reality Head-up Displays

Abstract

This dissertation proposes a pathway towards a safety-centric evaluation framework for augmented reality head-up displays (AR HUDs) in automotive contexts, addressing critical shortcomings in traditional testing methods and emphasizing the integration of human factors. Conducted through a three-phase, mixed-method research program, the present work critically evaluates the inadequacy of conventional glance-based metrics, highlights the safety risks posed by inattentional blindness induced by AR HUD interfaces, and fosters international, cross-sector dialogue to advance AR HUD evaluation practices. Study 1 examined sustained-glance thresholds, identifying three foundational components for safety assessment: the development of standardized reference tasks, the incorporation of ecologically valid driving environments, and the use of performance metrics directly tied to safety outcomes. Study 2 provided empirical evidence that drivers can fail to detect hazards even when looking at them—demonstrating that glance duration alone is insufficient to evaluate AR HUD safety. Study 3 facilitated a two-phase collaboration among academic researchers, industry professionals, and regulatory bodies, exposing persistent gaps between research, practice, and policy, and emphasizing the need for more effective knowledge translation. Building on these insights, this dissertation proposes a strategic roadmap for developing a robust safety evaluation framework for AR HUDs. Key components include the establishment of interdisciplinary, multi-stakeholder committees; the development of a validated evaluation toolkit tailored to AR HUD-specific challenges; the formulation of human-centered design guidelines; and the implementation of regulatory standards grounded in human factors. As the culmination of this research, we propose a safety-focused evaluation framework for AR HUDs that highlights the importance of situation awareness. This approach moves beyond traditional usability metrics to place driver safety at the center of design and assessment, especially in today's complex evolving driving environments.