Lane changing behavior is a common driving behavior of road vehicles. In order to analyze and predict the lane changing risk of vehicles in the target lane during side lane changing, this paper firstly processed the data and obtained the spatial position relationship of vehicles based on the actual data. Secondly, the possible collision time and collision probability are calculated, and the lane-changing collision risk calculation model is proposed, which is divided into three categories according to the severity. Thirdly, Light Gradient Boosting Machine (LightGBM) method is used to predict the three types of collision risks, and compared with Gradient Boosting Decision Tree(GBDT) and Support Vector Machine (SVM) methods, it is found that the prediction accuracy of LGBM method is more than 99%, which can effectively predict and identify the vehicle collision risk. Finally, the influencing factors are analyzed by importance ranking. It is found that the most influential factor of vehicle collision risk is the vehicle in front. Since the lateral vehicle will become the vehicle in front of the target vehicle after lane change, attention should be paid to maintaining a safe distance between the vehicles before and after the lane change.
Risk prediction of side-vehicle lane change collision
2022-10-01
662602 byte
Conference paper
Electronic Resource
English
Vehicle trajectory prediction and collision warning for lane change conditions
Taylor & Francis Verlag | 2024
|