Exploring the Impact of Interconnected External Interfaces in Autonomous Vehicles on Pedestrian Safety and Experience

Title of the Paper

Exploring the Impact of Connected External Interfaces of Autonomous Vehicles on Pedestrian Safety and Experience

Paper Information

• Field of Study: Human-Computer Interaction, Autonomous Vehicles, Pedestrian Traffic Safety
• Keywords: Autonomous Vehicles, External Communication, eHMIs, Vulnerable Road Users, Vehicle-Pedestrian Interaction, Scalability

Research Background and Issues

• Problems and Challenges:

  • With the proliferation of autonomous vehicles (AVs), addressing communication between AVs and pedestrians has become a critical issue, as traditional human communication methods (e.g., eye contact, gestures) are no longer effective in AV scenarios.
  • Standard external Human-Machine Interfaces (eHMIs) may cause information overload or conflicting signals for pedestrians in complex traffic scenarios, particularly in multi-lane and multi-vehicle environments.
  • Current solutions are mostly vehicle-centric, which may lead pedestrians to overlook surrounding vehicles when making decisions.

• Significance:

  • This research is crucial for the safety and fluidity of future urban traffic, especially in interactions between pedestrians and autonomous technologies, which have broad implications for urban areas.

• Research Motivation and Related Work:

  • Building on discussions in the literature about potential extensions of eHMIs and multi-vehicle networking, the concept of "connected eHMIs" is proposed, where multiple autonomous vehicles collaborate via networked eHMIs to output consistent and clear signals.
  • Existing discussions on connected eHMIs remain theoretical, lacking empirical studies on visual interaction systems.

Solution

• Proposed Solution:

  • Design and validate a visual-based connected eHMI system aimed at providing pedestrians with unified information about multi-lane traffic conditions.
  • Evaluate connected eHMIs using VR (Virtual Reality) simulations and experimental methods, comparing them to conditions with no eHMI and non-connected eHMI.

• Innovations:

  • Introduction of a red-green coding system: red crosswalks indicate caution, while green crosswalks signify complete safety.
  • Combination of single-vehicle communication and centralized communication: only one autonomous vehicle projects the crosswalk to reduce visual clutter, while other vehicles use synchronized light bands to reinforce signal consistency.
  • Integration of Wi-Fi icons to help pedestrians understand the connectivity between vehicles.

• Implementation Steps and Key Technologies:

  1. Design three interface scenarios (no eHMI, non-connected eHMI, and connected eHMI).
  2. Conduct experiments in VR simulations involving two-lane traffic, recording and analyzing pedestrian behavior, safety, and subjective experience.
  3. Measure workload (NASA-TLX scale), trust, and perceived safety using both objective and subjective indicators.

Research Findings

• Experimental Results and Discoveries:

  • Pedestrian Safety:
    • Connected eHMIs enhanced pedestrians' perception of safety, whereas scenarios without eHMIs induced more uncertainty.
    • In terms of actual collision metrics, connected eHMIs performed similarly to non-connected eHMIs but did not significantly outperform no eHMI; non-connected eHMIs may lead to signal misinterpretation (e.g., mistaking a green light for complete road safety).
  • Workload:
    • The three eHMI conditions did not show significant differences in their impact on pedestrian workload; however, the red-green signals and Wi-Fi icons in connected eHMIs caused cognitive confusion for some participants.
  • Trust:
    • Compared to no eHMI, both connected and non-connected eHMIs generally increased trust, but participants' understanding of specific designs, such as Wi-Fi icons, significantly influenced their trust in connected eHMIs.

• Comparative Advantages:

  • Compared to non-connected eHMIs, connected eHMIs were more effective in reducing the risk of signal misinterpretation.
  • The cautious prompts of red signals showed strong performance in enhancing pedestrian alertness.

• Limitations and Future Directions:

  1. Limitations:
     • VR experiments may not fully reflect real-world street environments.
     • Certain design elements (e.g., red crosswalks and Wi-Fi icons) lacked cognitive consistency.
  2. Future Directions:
     • Explore more intuitive and consistent signal languages (e.g., optimized colors, textual indicators).
     • Test eHMIs in real-world mixed traffic environments (autonomous and manual vehicles).
     • Investigate the positive role of education in improving pedestrians' understanding of eHMIs.

Conclusion

This study addresses some of the contradictions in signal transmission during pedestrian-autonomous vehicle interactions, proposing a new design for connected eHMIs and conducting systematic empirical tests. The findings demonstrate the potential of this design in enhancing perceived safety and reducing signal misinterpretation. However, its effectiveness and reliability in complex traffic conditions still require validation through larger-scale and more realistic scenarios.