Probabilistic Vehicle Trajectory Prediction Based on LSTM Encoder-Decoder and Attention Mechanism

In order to realize driving safety in highway scenarios, autonomous vehicles need to predict and reason about the driving intentions and motion trajectories of surrounding target vehicles in the near feature. Essentially, trajectory prediction of target vehicles can be viewed as a typical time series generation problem, which predicts the future trajectory of the vehicle through analyzing the input of historical trajectory information or its control signals. In actual traffic scenarios, the movement between vehicles is a process of mutual game and cooperation, namely the future trajectory of a vehicle is not only related to its own historical trajectory, but also to surrounding vehicles motion. However, different surrounding traffic participants have different influence on the target vehicle, and the future motion of the vehicle is often affected by some specific surrounding traffic agents deeply. Inspired by the driver's selective attention mechanism, this paper proposes a vehicle trajectory prediction method based on the Long-short Term Memory (LSTM) Neural Network and attention mechanism, which can predict the location probability distribution of the target vehicle while considering the interaction between vehicles. We train and test our model on the NGSIM dataset. The results show that the proposed method has significantly improved prediction accuracy compared with other existing advanced models in short-term and long-term prediction. The Root of the Mean Squared Error (RMSE) of the target vehicles’ position at the 5s prediction time is reduced to 1.28 m. By analyzing the influential factors, we find that expansion of the spatial perception range can improve the trajectory prediction accuracy effectively. At the same time, this paper also presents quantitative and qualitative analysis of our model and verifies the prediction effect in typical scenarios.