TL;DR
This study uses a VR simulation to investigate how human-machine co-driving vehicle speed and time headway affect pedestrian crossing speed at uncontrolled mid-block road sections. The findings reveal that gender and trust in the vehicle significantly influence crossing speed, with time headway being a stronger predictor than vehicle speed, offering crucial insights for designing safer pedestrian crossing controls in the era of autonomous driving.
What Problem Does This Paper Address?
As urban environments become more pedestrian-friendly and autonomous vehicle technology advances, the interaction between pedestrians and these new vehicles at uncontrolled crossing points becomes a critical safety concern. Traditional traffic safety research has focused on human-driven vehicles, but the transition phase of human-machine co-driving (L2/L3 automation) presents unique challenges. Pedestrians may struggle to accurately predict the behavior of these vehicles, and their trust levels can vary significantly, potentially leading to altered crossing speeds and increased accident risks. This paper aims to fill the gap in understanding how these specific factors—vehicle speed and time headway of co-driving vehicles—influence pedestrian crossing behavior, particularly at mid-block, non-signalized crossings.
Methods
The researchers employed a Virtual Reality (VR)-based simulation study to investigate pedestrian crossing behavior. A total of 30 college students participated in the experiment. They were tasked with completing five street-crossing trials in a simulated environment featuring human-machine co-driving vehicles. The simulation allowed for precise control and manipulation of two key variables: vehicle speed (five levels: 20, 30, 40, 50, 60 km/h) and time headway (five levels: 3, 4, 5, 6, 7 seconds). Participants' crossing speeds were recorded using a walking simulation apparatus (KAT Walk Mini S system) and Scanner Studio 1.6 software. Beyond the simulation, participants also completed a demographic questionnaire that included their level of trust in human-machine co-driving vehicles. Statistical analyses, including independent t-tests, partial correlations, and ANOVA, were used to analyze the collected data and identify significant relationships between the independent variables (gender, trust, time headway, vehicle speed) and the dependent variable (pedestrian crossing speed).
Datasets
Not discussed in the paper. The study utilized a VR simulation environment and collected data from 30 participants.
Key Insights and Ideas
- Trust is a double-edged sword: Counterintuitively, increased trust in human-machine co-driving vehicles led to decreased crossing speeds. This suggests that while trust is essential for adoption, lower trust might prompt pedestrians to cross faster to mitigate perceived risks, or conversely, higher trust might lead to more cautious, slower crossings.
- Time headway reigns supreme: The study found that the time headway of the approaching co-driving vehicle had a stronger correlation with pedestrian crossing speed than the vehicle's actual speed. This highlights the importance of the perceived "gap" or available time for crossing, rather than just the speed itself.
- Gender matters: Male pedestrians consistently crossed faster than female pedestrians, aligning with previous research on traditional traffic environments. This points to potential differences in risk-taking behavior or physical capabilities that need consideration in safety design.
- Co-driving vehicles are not just cars: The study emphasizes that the unique characteristics of human-machine co-driving vehicles (e.g., their automation status, potential for unpredictable behavior) introduce new dynamics to pedestrian interactions, distinct from conventional vehicles.
Key Results
- Gender Differences: Male pedestrians had an average crossing speed of 1.489 m/s, significantly higher than female pedestrians at 1.291 m/s (p < 0.001).
- Trust Impact: A significant negative correlation (r = -0.175, p < 0.001) was found between trust in human-machine co-driving vehicles and pedestrian crossing speed. As trust increased, crossing speed decreased. Specifically, pedestrians who distrusted the vehicles crossed significantly faster than those who trusted them.
- Vehicle Speed: Pedestrian crossing speed significantly increased with vehicle speed (r = 0.093, p = 0.011). For example, crossing speeds were significantly different at 20 km/h versus 50 km/h and 60 km/h.
- Time Headway: Pedestrian crossing speed significantly decreased with increasing time headway (r = -0.339, p < 0.001). This effect was more pronounced than that of vehicle speed.
- Relative Influence: Time headway had a stronger correlation with pedestrian crossing speed (r = -0.339) than vehicle speed (r = 0.093).
Limitations and Future Work
The authors acknowledge several limitations:
- Scenario Specificity: The study focused only on uncontrolled mid-block road sections. Future work should explore pedestrian behavior at intersections, which present different traffic flow dynamics and road attributes.
- Participant Demographics: The study primarily involved young college students. Future research should investigate the influence of different age groups on pedestrian crossing speed, as age is a known factor in cognitive and physical abilities relevant to crossing.
- Simulation vs. Reality: While VR simulations offer control, they may not perfectly replicate real-world complexities. The authors suggest conducting real vehicle experiments to validate the findings and enhance their engineering application value.
The findings provide a strong foundation for developing more effective pedestrian crossing control measures tailored to the evolving landscape of human-machine co-driving vehicles.
Summary updated