考虑驾驶意图的车辆换道轨迹预测研究

Research on vehicle lane change trajectory prediction considering driving intent

换道是导致交通事故的主要行为之一,对车辆的换道行为进行分析并预测换道轨迹有助于减少事故的发生。为了提高车辆换道轨迹的预测精度,提出一种考虑驾驶意图的车辆换道轨迹预测方法。利用余弦角相似度进行谱聚类的方法区分直行轨迹与换道轨迹,选取换道轨迹中12个特征参数,利用随机森林进行换道意图的辨识,最后搭建LSTM模型并考虑换道意图进行车辆轨迹预测。通过HighD数据集进行验证,结果表明,该方法能够有效提高换道轨迹预测的准确性。

Lane changing behavior is one of the main behaviors that lead to traffic accidents,and analyzing the lane changing behavior of vehicles and predicting their trajectory can help reduce the occurrence of accidents.In order to improve the prediction accuracy of vehicle lane change trajectory,a vehicle lane change trajectory prediction method considering driving intention was proposed.Firstly,the cosine angle similarity for spectral clustering was used to distinguish the straight trajectory from the lane change trajectory,12 feature parameters in the lane change trajectory were selected to identify the lane change intention using random forest,and finally the LSTM model was built and the lane change intention was considered for vehicle trajectory prediction.The identification was performed by HighD dataset,and the results showed that the method could effectively improve the accuracy of lane change trajectory prediction.