Vehicle Trajectory Prediction Based on GAT and LSTM Networks in Urban Environments

Abstract

Vehicle trajectory prediction plays a critical role before the decision planning of autonomous vehicles in complex and dynamic traffic environments. It helps autonomous vehicles better understand the traffic environments and ensure safe and efficient tasks. In this study, a hierarchical trajectory prediction method is proposed. The graph attention network (GAT) model was selected to estimate the interactions of surrounding vehicles. Considering the behaviour of surrounding agents, the future trajectory of the target vehicle is predicted based on the long short-term memory network (LSTM). The model has been validated in real traffic environments. By comparing the accuracy and real-time performance of target vehicle trajectory prediction, the proposed model is superior to the traditional single trajectory prediction model. The results of this study will provide new modelling ideas and a theoretical basis for the vehicle trajectory prediction in urban traffic environments.