ST-AGNN: Spatial-Temporal Attention Graph Neural Network for Pedestrian Trajectory Prediction

Abstract

Accurate and fast prediction of pedestrian trajectory is very important. However, the interaction between pedestrians is complex, pedestrians are affected not only by their own motion but also by the neighboring pedestrians. Some methods do not extract the pedestrian motion features completely. It is easy to ignore a few of important features of pedestrians. The Spatial-Temporal Attention Graph Convolutional Network (ST-AGNN) for pedestrian trajectory prediction is proposed in this paper. ST-AGNN model can extract the spatial interaction features of each time step and stack the spatial features to obtain the spatial-temporal features. It can predict the future trajectory by using Temporal Convolutional Network (TCN). We evaluate the average displacement error (ADE) and final displacement error (FDE) of the proposed method on ETH and UCY datasets. The experimental results show that the proposed method is effective in accuracy and efficiency.