From Awareness to Intent: Mitigating Silent Driving System Failures through Prospective Situation Awareness Enhancing Interfaces

Abstract: Silent automation failures, where a system fails to detect a hazard without warning, pose a critical safety challenge for partially automated vehicles. While research has mostly focused on takeover requests, how to support a driver in silent failure remains underexplored. We conducted a multi-modal driving simulator study with 48 participants to investigate how different Prospective Situation Awareness Enhancement (PSAE) interfaces, delivered via augmented reality head-up display, affect takeover performance. By integrating behavioral, subjective psychological, and physiological data, our analysis suggests that situational awareness (SA) serves as an important moderating factor through which PSAE interfaces improve takeover performance. Further, we found that providing perceptual cues was most effective in enhancing SA, while communicating system intent was superior for building trust. Finally, we identified a potential correlate of SA in the neuroactivity. Overall, this paper contributes to understanding how transparency-oriented interfaces may support drivers and provides design insights into HMI design for silent failures.

• The paper demonstrates that AR-based PSAE interfaces significantly improve driver situation awareness during silent ADS failures.
• Methodology employs a mixed factorial simulator study with EEG, EMG, and behavioral metrics across varied lighting and hazard conditions.
• Results reveal that enhanced SA mediates takeover performance, underscoring the need for clear perceptual cues to boost trust and reduce cognitive overload.

Mitigating Silent Driving System Failures: Impact of Prospective Situation Awareness Enhancement Interfaces

Introduction and Motivation

Silent automation failures in SAE Level 2 vehicles—where the automated driving system (ADS) fails to detect road hazards without issuing a takeover request (TOR)—present a substantial safety risk due to the lack of explicit driver warning. Research has traditionally emphasized driver response to explicit TORs, neglecting the highly prevalent and safety-critical silent failure scenario. This paper proposes and evaluates Prospective Situation Awareness Enhancement (PSAE) interfaces delivered via augmented reality (AR) head-up displays (HUD), designed to transparently communicate ADS perception and intent, in order to support drivers during silent failures.

The study systematically interrogates the effects of PSAE interface design, environmental condition (day/night lighting), and hazard visibility (visible/invisible) on driver psychological states (situation awareness, perceived safety, trust), neurophysiological indicators (EEG/EMG), and takeover performance.

Figure 1: Experimental manipulation framework illustrating PSAE and lighting conditions in the simulator.

Experimental Design and Methodology

A mixed factorial design was employed with one within-subject factor (hazard visibility) and two between-subject factors (PSAE interface and lighting). Forty-eight experienced drivers each encountered six silent-failure events (three visible, three invisible hazards) in a high-fidelity, six-degree-of-freedom driving simulator—immersive urban environments were recreated in CARLA, with precise event timing and AR overlays manipulated using a Wizard-of-Oz protocol.

Three PSAE visualization conditions were defined: (1) Environment Perception (EP): bounding box overlays of detected objects; (2) Planned Maneuver (PM): ADS-planned trajectory visualization; (3) Combined (EP+PM): simultaneous display of detected objects and intended maneuvers. A Baseline condition (no PSAE) served as control. Silent failures were implemented by omitting hazard overlays (perceptual misses) or portraying trajectories that conflicted with hazards. No alarms or TORs were issued during these failures.

Driver psychological states were measured via SART, trust scales, and perceived safety questionnaires post-event. EEG (32-channel, 500 Hz) and EMG (1000 Hz, hand/leg sites) signals were synchronized to behavioral logs for multilayered neuroergonomic analysis. Takeover performance metrics included lead time and success rate.

Figure 2: (a) Simulator hardware and (b) closed-loop urban driving route setup.

Theoretical Framework and Statistical Approach

The analytical framework followed hierarchical information processing and situation awareness theory, positing sequential influence from environmental/interface variables → psychological states → physiological indicators → behavioral performance. Regression analyses (generalized linear and mixed models), supplemented by structural equation path modeling (SEM) with mediation analysis, were used to test direct, indirect, and interaction effects, controlling for individual and repeated measures.

Figure 3: Theoretical causal model outlining environment, psychological, physiological, and behavioral layers.

Key Findings

Effects of PSAE and Environmental Factors

Regression results confirmed that both PSAE interface and lighting conditions significantly influence driver psychological states and preparation:

• SA was significantly higher with EP interfaces compared to Baseline or Combined EP+PM.
• PM and EP+PM interfaces significantly increased driver trust.
• Nighttime driving led to lower trust but higher muscle preparation and SA scores—indicating a vigilance effect.
• Scenario type (visible hazards) strongly predicted perceived safety and trust.

Interaction analyses showed that PSAE effects on neural activity (EEG frontal theta power ratio and activation time) were moderated by situation awareness.

Figure 4: Main effects of lighting, PSAE, and scenario type on SA, perceived safety, trust, and neurophysiological measures.

Psychological–Physiological Linkages

Higher SA was robustly associated with decreased parietal alpha power ratio (i.e., alpha suppression), earlier parietal alpha activation time, and increased frontal theta power—neural signatures of heightened cognitive engagement and conflict monitoring. However, the temporal coupling between cortical detection (EEG) and motor initiation (EMG) was not tight: higher SA was linked to longer EEG-EMG time lag, suggesting cognitive processing does not always immediately translate to physical action in silent-failure scenarios.

Figure 5: Interaction effects between SA and PSAE on frontal theta power ratio and activation time, with regression fits and correlation analysis.

Mediation Mechanism: PSAE → SA → Performance

Path analysis established that the effect of PSAE interfaces on takeover success was fully mediated by SA:

• Significant indirect effects were detected for all PSAE conditions vs. Baseline (EP: 0.431, PM: 0.327, EP+PM: 0.209), with no significant direct effects on performance.
• For takeover lead time, EP yielded a significant mediated effect; PM showed trend-level mediation.
• Simple PSAE → SA → Physiological chains did not fit the data, suggesting complexity or insufficient power.

Implications for HMI Design and Theoretical Advancement

The results reinforce the centrality of situation awareness in successful takeover during silent ADS failures. PSAE interfaces primarily enhance takeover performance through cognitive pathway improvements, rather than direct behavioral or physiological modulation; clear perceptual cues (EP) are optimal for SA, while maneuver information fosters trust—combined information does not guarantee additive benefits and may induce overload.

The consistent link between SA and parietal alpha suppression offers a neurophysiological marker for real-time driver monitoring, supporting the future integration of adaptive, context-aware HMIs leveraging neural metrics. The "night-time vigilance effect" and interaction findings suggest dynamic adaptation of HMI content based on driver states and environmental context is required.

Limitations

The sample size per PSAE group was constrained by the need to avoid carryover, leading to limited power in interaction and moderation analyses. Simulator conditions, while realistic, may not fully capture real-world emotional and physiological risk. Path mediation of physiological variables was not supported, indicating further empirical and methodological refinement is needed, including finer temporal alignment and richer scenario diversity.

Conclusion

This study provides a rigorous empirical and neuroergonomic foundation for the role of transparency-oriented PSAE interfaces in mitigating silent ADS failures. By integrating AR-HUD overlays and multi-modal state measurement, the results delineate situation awareness as a pivotal mediator between interface design and takeover performance. The findings urge cautious, context-sensitive HMI design that foregrounds perceptual clarity, minimizes cognitive overload, and adapts to both environmental and driver state. The demonstrated neural correlates of SA substantiate future directions in physiological monitoring and adaptive interfaces for enhanced safety in automated driving systems (2604.18449).