No abstract

No abstract

The focus of this paper is the crash risk assessment of off-ramps in Xi’an. The time-to-collision (TTC) is used for the measurement and cross-comparison of the crash risk of each location. Five sites from the urban expressway in Xi’an were selected to explore the TTC distribution. An unmanned aerial vehicle and a camera were used to collect traffic flow data for 20 min at each site. The parameters, including speed, deceleration rate, truck percentage, traffic volume, and vehicle trajectories, were extracted from video images. The TTCs were calculated for each vehicle. The Gaussian mixture model (GMM) was proposed to predict the TTC probability density functions (PDFs) and cumulative density functions (CDFs) for five sites. The Kolmogorov–Smirnov (K-S) test indicated that the samples followed the estimated GMM distribution. The relationship between the crash risk level and influencing factors was studied by an ordinal logistic regression model and a naive Bayesian model. The results showed that the naive Bayesian model had an accuracy of 86.71%, while the ordinal logistic regression model had an accuracy of 84.81%. The naive Bayesian model outperformed the ordinal logistic regression model, and it could be applied to the real-time collision warning system.

No abstract

The traffic environment at the exit of the urban expressway is complex, and vehicle lane-changing behavior occurs frequently, making it prone to traffic conflict and congestion. To study the traffic conditions at the exit of the urban expressway and improve the road operation capacity, this paper analyzes the characteristics of lane-changing behaviors at the exit, adds driving style into the influencing factors of lane-changing, and recognizes one’s lane-changing intention based on driving data. A UAV (unmanned aerial vehicle) is used to collect the natural driving track data of the urban expressway diverge area, the track segments of vehicle lane-changing that meet the standards are extracted, and 374 lane-changing segments are obtained. K-means++ is used to cluster the driving style of the lane-changing segments which is grouped into three clusters, corresponding to “ordinary”, “radical”, and “conservative”. Through the random forest model used to identify and predict driving style, the accuracy reaches 93%. Considering the characteristics of a single time point and the characteristics of the historical time window, XGBoost, LightGBM, and the Stacking fusion model are established to recognize one’s lane-changing intention. The results show that the models can well recognize the lane-changing intention of drivers. The Stacking fusion model has the highest accuracy, while the LightGBM model takes less time; the model considering the characteristics of the historical time window performs better than the other one, which can better improve the prediction accuracy of lane-changing behavior.