Vehicle Trajectory Prediction in Roundabout Considering Real-Time Risk Feature: A Deep Neural Network Approach with Attention Mechanism

Vehicle trajectory prediction aims to estimate the future location of the vehicle based on its historical motion state. However, existing researches often ignore the impact of environmental factors on safe driving during the prediction process in roundabouts. Therefore, this study focuses on vehicle trajectory prediction in roundabouts and proposes a hybrid model (CNN-LSTM-Attention) that combines CNN and LSTM networks with an attention mechanism. This model predicts future trajectories by acquiring vehicle motion states and risk features. The proposed method is evaluated and compared with existing approaches on the publicly available rounD data set. The experimental results show that the proposed model significantly enhances trajectory prediction accuracy, with its output trajectories better fitting the actual vehicle trajectories.