BikeAR: Understanding Cyclists' Crossing Decision-Making at Uncontrolled Intersections using Augmented Reality

Title of the Paper

BikeAR: Understanding Cyclists’ Crossing Decision-Making at Uncontrolled Intersections using Augmented Reality

Paper Information

• Subject Area: Human-Computer Interaction and Traffic Safety
• Keywords: Augmented Reality, Bicycle Safety, Intersection Crossing Decision-Making, Mixed Reality, Urban Traffic

Research Background and Problem

• Problem or Challenge:
  • Cyclists are a vulnerable group in urban traffic, facing significant safety risks, especially at uncontrolled intersections.
  • Traditional bicycle assistance systems have technical limitations, such as the inability to dynamically convey information about occluded vehicles.
• Significance:
  • Uncontrolled intersections are high-risk areas for traffic accidents, and decision-making support for cyclists can significantly enhance their safety.
  • The increase in urban traffic flow and the future integration of autonomous vehicles make understanding cyclists’ crossing strategies more urgent.
• Research Motivation and Related Work:
  • Existing technologies, such as multimodal assistance systems, head-mounted displays, and projection interfaces, contribute to safety but often suffer from spatial separation of information, failing to effectively address the presentation of occluded vehicle information.
  • This study aims to provide an augmented reality method to support cyclists’ dynamic decision-making, overcoming the limitations of existing systems.

Solution

• Method or Solution:
  • Proposed two augmented reality visualization methods:
    1. X-ray: Displays occluded vehicles using "see-through" technology.
    2. Countdown: Displays the remaining safe crossing time at the intersection in the form of a timer.
  • Designed a safe experimental framework using augmented reality glasses and a simulated virtual environment.
• Innovations:
  • First application of augmented reality technology to evaluate cyclists’ decision-making processes, integrating visual information to improve traffic scene perception under occlusion conditions.
  • Proposed a user research method based on augmented reality simulation of real physical environments, supporting safe, immersive, and realistic experimental conditions.
• Implementation Steps and Key Technologies:
  • Participants ride indoors while wearing augmented reality glasses to observe a virtual urban traffic environment.
  • Built a modular virtual city scene using the Unity engine, including intersection configurations and dynamic traffic flow simulations.
  • Combined virtual experiments with physical cycling to enhance ecological validity.

Research Outcomes

• Specific Outcomes:
  • Provided empirical evaluation results of two augmented reality visualization techniques under different traffic densities:
    • X-ray visualization enhanced cyclists’ environmental awareness, helping users choose short gaps in traffic flow and reducing psychological stress.
    • Countdown provided a better overview of intersection traffic, increasing cyclists’ sense of safety.
  • Cyclists relied on these technologies for early decision-making, reducing crossing time.
• Advantages:
  • Compared to no assistance, augmented reality technology enabled cyclists to make faster decisions without increasing accident risk.
  • Countdown demonstrated a precise way to represent dynamic traffic gaps, while X-ray offered broader environmental awareness support.
• Experimental or Evaluation Results:
  • Cyclists tended to choose short gaps in sparse traffic but relied more on X-ray visualization for dynamic decision-making in dense traffic conditions.
  • X-ray was more effective in reducing cyclists’ psychological stress, while Countdown improved participants’ overall sense of safety at intersections.
  • Using the NASA Task Load Index to evaluate workload showed that X-ray had lower psychological burden in dense traffic scenarios.
• Limitations and Future Directions:
  • The virtual environment in the experiment was limited to visual simulation, lacking consideration of sound and other environmental factors.
  • Indoor space constraints led to adjustments in traffic flow parameters, not fully reflecting real cycling environments.
  • Proposed integrating visual technologies into existing helmets, traffic signal facilities, or Car2X technologies to expand application scenarios and improve system response speed.

Conclusion

This study demonstrated that augmented reality support tools effectively improved cyclists’ decision-making abilities at uncontrolled intersections and established a safe experimental framework for further research. Future work could focus on optimizing visual feedback, multimodal integration, and deployment in real-world scenarios to contribute to the development of smarter urban traffic systems.