This paper introduces a novel map-free trajectory prediction model, MFTraj, that leverages historical trajectory data and a dynamic geometric graph-based behavior-aware module to capture complex interactions in dynamic traffic scenarios without relying on high-definition maps.
CASPFormer is a novel motion prediction architecture that generates diverse, scene-consistent multi-modal trajectories from rasterized Bird's-Eye-View (BEV) images, without relying on expensive High-Definition maps.