STAG: A novel interaction-aware path prediction method based on Spatio-Temporal Attention Graphs for connected automated vehicles

Mozhgan Nasr Azadani, Azzedine Boukerche

Abstract: Highlights•A novel vehicle path prediction method is proposed based on Graph Neural Network.•A novel asymmetric interaction reasoning for the vehicle and its surrounding agents.•A new spatio-temporal attention mechanism to extract the trajectory representations.•STAG outperforms benchmarks models in terms of RMSE in highly structured highways.•Lower errors in terms of ADE/FDE in unsignalized intersections and roundabouts.