.png){kind=icon}
.png){kind=icon}
Return
Analyzing continuous lane-changing behavior using freeway drone dataset
.png){kind=icon}
PRE
AI
DOI：10.1080/15472450.2025.2531359.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Continuous lane-changing (CLC) events are relatively rare yet significantly impact traffic flow and safety. Analyzing CLC behavior is crucial for microsimulation and traffic management; however, existing studies rarely investigate its characteristics. This study proposes an automated method to extract 122 CLC events from the HighD dataset and analyzes the key characteristics of duration, mean speed, relative speed between the subject vehicle (SV) and the original lane’s preceding vehicle (OPV), safety, and impact on the target lane’s following vehicle (TFV). A Cox Proportional Hazards (CPH) model with time-varying covariates and coefficients models the lane line crossing duration. Safety is evaluated using time-to-collision (TTC) and safety rate, while impact is assessed through speed change rate and maximum deceleration. The main findings are: (1) duration varies from 2.76 s to 12.28 s, with CLC events crossing lane lines more quickly compared to single lane-changing (SLC) events, and the longitudinal speed of the vehicle greatly influencing the lane line crossing duration; (2) CLC events are more likely to occur when the OPV is farther from the SV; (3) The transition lane is the most dangerous of the three lanes. This study provides valuable insights for advanced driver assistance systems (ADAS) to effectively identify CLC behaviors and enhance their capacity to improve road safety. The results also have implications for the development of connected and automated vehicles (CAVs).
Keywords:
continuous lane-changing
duration model
lane-changing behavior
safety
Journal
IF:
3.8
Papers:
4.6K
Citations:
1.2W
Organization
View more organizations