A Comparative Study of Graph Neural Network Layer Selection for Interaction Modelling in Driving Trajectory Prediction

In three linesComparative study of 19 GNN layer types for trajectory prediction in autonomous driving. ARMA, Chebyshev, and topology-aware layers consistently outperform others. Sum-based aggregation, multi-head attention, and distance-weighted hops significantly improve prediction accuracy.

Summary generated by Claude — human-verified