Answers are grounded in this paper and cite their sources.
Vehicle trajectory prediction is essential for collision warning in cooperative driving. However, trajectory prediction becomes difficult because data availability varies across dynamic cooperative driving scenarios. Most previous methods are scenario-specific and have limited feature extraction capacity under uncertain inputs. To address this challenge, this paper proposes a local-global learning method for adaptive trajectory prediction (LGL-ATP). A local spatial-temporal feature extraction module is designed to compensate for insufficient local contextual representations, thus improving short-term prediction accuracy under limited perception scenarios. Additionally, global spatial interactions are modeled to automatically learn extensive traffic influences, and an intention-guided vehicle mask prioritizes processing of future trajectory-relevant areas, thus enhancing long-term prediction while suppressing redundant features. Experiments conducted on two types of cooperative driving scenarios demonstrate the adaptability of the proposed method compared to the state-of-the-art models.
No takes yet. Share an insight, caveat, or question.
Liu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: