Saliency-Based Attention Shifting: A Framework for Improving Driver Situational Awareness of Out-of-Label Hazards

Shleibik, Yousra; Sinclair, Jordan; Haring, Kerstin · 2025 · arXiv

archive: archived pipeline: cataloged verified

Get this paper ↗ (full text — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the critical safety challenge of driver situational awareness during takeover requests (TORs) in semi-autonomous vehicles, specifically when encountering "out-of-label" hazards that the autonomous system cannot identify or handle independently. The authors argue that current human-vehicle interaction systems are insufficient because drivers often engage in non-driving related tasks (NDRTs) or suffer from target fixation, where they hyperfocus on a single object rather than scanning the broader environment. This narrow focus reduces the driver’s ability to execute safe evasive maneuvers. The research is motivated by the need to bridge the gap between automated vehicle capabilities and human decision-making, ensuring a seamless transition of control even when the driver is distracted or lacks immediate context about the hazard. To address this, the authors propose a conceptual framework that integrates real-time gaze tracking, context-aware saliency analysis, and multimodal alerts. The system utilizes an inward-facing camera to track the driver’s gaze vector and fixation points. When an anomaly detection algorithm identifies an unlabeled hazard, the system generates a filtered saliency map to identify high-value regions relevant to the hazard and potential secondary risks. Using this data, the system calculates a trajectory from the driver’s current fixation point to the hazard location, incorporating intermediate waypoints for critical environmental features. This trajectory is then used to guide the driver’s attention through a Head-Up Display (HUD) and synchronized audio alerts. The HUD presents sequential visual cues, such as animated arrows, pulsing indicators, or color-coded overlays, which adapt based on hazard urgency. Audio alerts, ranging from low-level tones to urgent beeps, are timed to reinforce these visual shifts and break target fixation. The paper does not present empirical results from a conducted experiment but rather outlines the theoretical design and operational logic of the proposed system. It details how the fusion of gaze data and saliency mapping allows for the dynamic generation of attention-guiding trajectories. The framework specifies that visual cues are projected directly into the driver’s forward field of view to minimize distraction, while audio cues leverage sound’s ability to capture immediate awareness. The authors illustrate the system’s workflow through diagrams showing how gaze trajectories are generated and how HUD cues are displayed to redirect attention from a distraction point to the hazard and related safety zones. The significance of this work lies in its potential to mitigate the risks associated with target fixation and delayed reaction times during autonomous vehicle takeovers. By proactively guiding driver attention to the most critical aspects of the scene, the framework aims to reduce the time required to regain situational awareness and improve the safety of human-vehicle collaboration. The authors conclude that this approach offers a promising avenue for enhancing driver readiness in uncertain conditions. They note that future work will involve developing a functional prototype and conducting comprehensive user studies in high-fidelity simulation environments to validate the system’s effectiveness in handling out-of-label hazards.

Key finding

A gaze-aware framework using filtered saliency maps and synchronized visual/auditory cues is proposed to break target fixation and accelerate driver situational-awareness recovery during takeovers from out-of-label hazards.

Methodology

theoretical

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via discover_arxiv on 2026-05-04 (5 acquisition events logged).

Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• situational awareness
• gaze based attention detection
• peripheral attention
• visual
• automation surprise
• hud ar windshield

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Methodological Resource: tool software
• Theoretical Contribution: conceptual framework, theory or model