PFR-HiVT: Enhancing Multi-Agent Trajectory Prediction with Progressive Feature Refinement
Open Access
Abstract
Multi-agent trajectory prediction is essential for autonomous driving systems, as its performance heavily depends on the quality of feature representations. This paper proposes PFR-HiVT, a lightweight and effective approach for multi-agent trajectory prediction, and evaluates it on the Argoverse 1.1 motion forecasting dataset. Although existing methods such as the Hierarchical Vector Transformer (HiVT) have achieved strong performance, they still exhibit limitations in feature extraction and feature transition across different stages of the network. To address these limitations, a collaborative feature enhancement framework is introduced, consisting of two encoder-side modules and a Progressive Feature Refinement Global Interactor (PFR-Global Interactor). Specifically, the Feature Enhancement Module (FEM) and the Attention Enhancement Module (AEM) are employed to refine local spatiotemporal features before global interaction. In addition, the PFR-Global Interactor integrates three lightweight components—the Simple Feature Refinement Module (SFR), the Lightweight Gate Module (LG), and the Residual Connection Module (RC)—to progressively refine globally interacted features prior to trajectory decoding. All proposed modules adopt lightweight designs, introducing only 230.5 k additional parameters (approximately 8.7% of the total parameters of HiVT-128). Experiments on the Argoverse 1.1 dataset show that PFR-HiVT achieves a minADE of 0.703, a minFDE of 1.041, and an MR of 0.112, outperforming the baseline HiVT model. Ablation studies further validate the effectiveness and synergy of the proposed modules.Keywords
Funding Information
- Inner Mongolia Education Department Project (JY20230118)
- Key R&D and Achievement Transformation Program in the Public Welfare Field during the 14th Five-Year Plan of Inner Mongolia (2023YFSH0003)
- Scientific Research Project of Colleges and Universities in Inner Mongolia Autonomous Region (NJZY22387)
- Scientific Research Project of Inner Mongolia University of Technology (ZY202018)
This publication has 25 references indexed in Scilit:
- Efficient multi-target vehicle trajectory prediction based on multi-scale graph convolutionPattern Analysis and Applications, 2024
- MTR++: Multi-Agent Motion Prediction With Symmetric Scene Modeling and Guided Intention QueryingIEEE Transactions on Pattern Analysis and Machine Intelligence, 2024
- Research on vehicle trajectory fusion prediction based on physical model and driving intention recognitionProceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, 2023
- Bootstrap Motion Forecasting With Self-Consistent Constraints2023
- Query-Centric Trajectory Prediction2023
- Wayformer: Motion Forecasting via Simple & Efficient Attention NetworksarXiv, 2022
- HiVT: Hierarchical Vector Transformer for Multi-Agent Motion Prediction2022
- GSAN: Graph Self-Attention Network for Learning Spatial–Temporal Interaction Representation in Autonomous DrivingIEEE Internet of Things Journal, 2021
- TPCN: Temporal Point Cloud Networks for Motion Forecasting2021
- Learning Motion Patterns of People for Compliant Robot MotionThe International Journal of Robotics Research, 2005