Can You Hazard a Guess?: Evaluating the Effect of Augmented Reality Cues on Driver Hazard Prediction

Title of the Paper

Can You Hazard a Guess? Evaluating the Effects of Augmented Reality Cues on Driver Hazard Prediction

Paper Information

• Domain: Human-Computer Interaction and Autonomous Driving Technology (Application of AR in vehicle perception and driving)
• Keywords: Autonomous vehicles, augmented reality, in-vehicle displays, takeover request, cues, attention, situational awareness

Research Background and Problem

• Problem or Challenge: Autonomous vehicles allow users to engage in non-driving-related tasks (NDRTs), but these tasks reduce drivers' critical situational awareness, potentially causing safety issues when a takeover request (TOR) arises. Existing traditional in-vehicle displays (Head-Down Display, HDD) further weaken drivers' situational awareness.

• Research Importance: The development of autonomous driving technology requires addressing the issue of driver-vehicle collaborative control, ensuring that drivers can maintain basic attention and prediction capabilities regarding road conditions even while engaging in non-driving tasks, thereby enhancing the safety of autonomous driving.

• Research Motivation: Augmented Reality (AR) offers the possibility of presenting non-driving tasks via Heads-Up Display (HUD), which may reduce the decline in situational awareness caused by task distractions. However, it remains unclear whether AR displays can optimize drivers' situational awareness and whether the inclusion of additional cue information is effective.

Solution

• Research Method or Solution: The authors designed two experiments to compare the impact of different display methods (HUD, HDD) on drivers' situational awareness during non-driving tasks and to explore the effects of dynamic attentional cues in AR HUD design.

• Innovations:

  1. Investigated how dynamic cues help drivers identify hazards and improve situational awareness.
  2. Conducted an in-depth comparison of HUD and HDD in the context of non-driving tasks.
  3. Proposed guidelines for AR HUD design in in-vehicle systems.

• Implementation Steps and Techniques:

  • Experimental Design:
    1. The first experiment tested the differences in situational awareness between AR displays (static and dynamic HUD) and traditional HDD during non-driving tasks.
    2. The second experiment focused on the collaborative effects of more complex non-driving tasks (simulated phone input tasks) and dynamic AR cues.
  • Technologies Used: Augmented Reality, HoloLens 2 mixed reality devices, driving simulation systems.

Research Findings

• Specific Findings:

  1. All forms of non-driving tasks weakened drivers' situational awareness and hazard prediction ability, showing a decline compared to the baseline condition of focusing solely on driving tasks.
  2. AR HUD did not show significant advantages over traditional HDD in situational awareness, except when dynamic cues were included.
  3. The workload of non-driving tasks significantly affected the effectiveness of cues, with more complex tasks reducing the effectiveness of dynamic cues.

• Comparison with Existing Solutions:

  • Simple HUD displays may not necessarily outperform traditional HDD.
  • Dynamic cues can improve situational awareness in certain driving scenarios but require further optimization in design.

• Experimental or Evaluation Results:

  1. Under AR HUD conditions with dynamic cues, drivers detected hazards more quickly, though this effect diminished under complex task conditions.
  2. NASA TLX workload surveys indicated that cue design is particularly important for high-workload tasks.
  3. The experiments confirmed that dynamic cues using color and position changes are potentially effective for enhancing driver attention.

• Limitations and Future Directions:

  1. The experiments were primarily based on short-term driving video simulations and did not address fatigue or distraction effects in long-term driving scenarios.
  2. Interface design only considered visual cues; future research could explore multimodal (visual + auditory) cues for collaborative effects on situational awareness.
  3. Cue design needs further optimization to minimize interference with non-driving task operations in complex task environments.

Summary and Recommendations

1. Simple HUD displays have limited effectiveness in enhancing situational awareness during driving; only dynamic cue designs can help improve attention related to driving.
2. In AR HUD development, it is recommended to design dynamic cues and avoid direct conflicts with non-driving tasks to reduce user distraction.
3. Integration of non-driving tasks and cues should consider task workload, optimizing interface experience to ensure driving safety.