FIF: future interaction forecasted for multi-agent trajectory prediction

Yang Cao

doi:10.1016/j.trc.2025.105190

FIF: future interaction forecasted for multi-agent trajectory prediction

2025-06-08

0

PRE

AI

DOI：10.1016/j.trc.2025.105190

Original

Original request for help

Save

Abstract

En 中文

• A future interaction-focused model is proposed to enhance prediction accuracy and efficiency. • A hierarchical fusion framework asynchronously updates token features for better attention use. • NETA-TPred, a large-scale dataset from Shanghai, China, captures complex urban traffic scenes. • The lightweight FIF model runs at 16 fps with 80 MiB memory, enabling real-time deployment.

FIF: future interaction forecasted for multi-agent trajectory prediction

Abstract

Journal

Organization