赵放,蔡英

• 1:北京信息科技大学计算机学院

摘要(Abstract)：

传统车辆轨迹预测方法依赖集中式训练，难以满足车联网环境下数据隐私保护和分布式协同的需求。现有基于联邦学习的轨迹预测方法在时空特征建模上存在不足，往往忽略车辆间的空间交互关系，对时空特征的挖掘不充分，且缺乏高效的模型聚合策略。对此，结合多模态时空图注意力网络（multimodal spatio-temporal graph attention network, MSTGAT）和双重权重联邦聚合（dual-weight federated aggregation, DWFA）算法，提出了一种基于联邦学习的时空信息融合轨迹预测方法 DWFA-MSTGAT。MSTGAT结合多尺度时间卷积和改进的Transformer编码器，以捕捉车辆轨迹不同尺度的时序依赖关系，利用时空图网络建模车辆间的动态交互，实现对时空特征的有效表征；DWFA综合考虑客户端模型性能和数据量，优化全局模型更新，提升模型在异构数据环境下的泛化能力。实验结果表明，DWFA-MSTGAT在保护数据隐私的同时，提高了轨迹预测的准确性，优于现有集中式和联邦学习方法。

关键词(KeyWords)： 车辆轨迹预测;深度学习;联邦学习;Transformer;图注意力网络

Abstract:

Keywords:

基金项目(Foundation): 国家自然科学基金项目（61672106）

作者(Author): 赵放,蔡英

DOI: 10.16508/j.cnki.11-5866/n.2025.05.003

参考文献(References)：

• [1] LEFèVRE S,VASQUEZ D,LAUGIER C. A survey on motion prediction and risk assessment for intelligent vehicles[J].ROBOMECH Journal,2014,1:1.
• [2] WANG Z B,GUO J Y,HU Z M,et al. Lane transformer:a highefficiency trajectory prediction model[J]. IEEE Open Journal of Intelligent Transportation Systems,2023,4:2-13.
• [3]王昀，蔡英，范艳芳，等.基于GAT和Transformer的车辆行为预测[J].小型微型计算机系统，2025,46(1):23-32.WANG Y,CAI Y,FAN Y F,et al. Vehicle behavior prediction based on GAT and Transformers[J]. Journal of Chinese Computer Systems,2025,46(1):23-32.(in Chinese)
• [4] MCMAHAN H B, MOORE E, RAMAGE D, et al.Communication-efficient learning of deep networks from decentralized data[C]//Proceedings of Machine Learning Research. Brookline, MA, USA:Microtome Publishing,2017, 54:1273-1282.
• [5] ALAHI A,GOEL K,RAMANATHAN V,et al. Social LSTM:human trajectory prediction in crowded spaces[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition(CVPR). New York, USA:IEEE,2016:961-971.
• [6] ZHANG X C, ANGELOUDIS P,DEMIRIS Y. Dual-branch spatio-temporal graph neural networks for pedestrian trajectory prediction[J]. Pattern Recognition,2023,142:109633.
• [7] WANG J Q, LIU K,LI H T. LSTM-based graph attention network for vehicle trajectory prediction[J]. Computer Networks,2024,248:110477.
• [8] SCHUBERT R,RICHTER E,WANIELIK G. Comparison and evaluation of advanced motion models for vehicle tracking[C]//2008 11th International Conference on Information Fusion. New York, USA:IEEE,2008:1-6.
• [9] DEO N,RANGESH A,TRIVEDI M M. How would surround vehicles move?A unified framework for maneuver classification and motion prediction[J]. IEEE Transactions on Intelligent Vehicles,2018,3(2):129-140.
• [10] ZYNER A,WORRALL S,NEBOT E. A recurrent neural network solution for predicting driver intention at unsignalized intersections[J]. IEEE Robotics and Automation Letters,2018,3(3):1759-1764.
• [11] DEO N,TRIVEDI M M. Convolutional social pooling for vehicle trajectory prediction[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops(CVPRW). New York, USA:IEEE,2018:1549-1557.
• [12] VASWANI A,SHAZEER N,PARMAR N,et al. Attention is all you need[C]//Advances in Neural Information Processing Systems. La Jolla, California, USA:NIPS, 2017,30:1-15.
• [13] GENG M S,LI J Y,XIA Y J,et al. A physics-informed Transformer model for vehicle trajectory prediction on highways[J]. Transportation Research Part C:Emerging Technologies,2023,154:104272.
• [14] QUINTANAR A,FERNáNDEZ-LLORCA D,PARRA I,et al.Predicting vehicles trajectories in urban scenarios with transformer networks and augmented information[C]//2021IEEE Intelligent Vehicles Symposium(IV). New York, USA:IEEE,2021:1051-1056.
• [15] LI X, YING X W, CHUAH M C. GRIP:graph-based interaction-aware trajectory prediction[C]//2019 IEEE Intelligent Transportation Systems Conference(ITSC). New York, USA:IEEE, 2019:3960-3966.
• [16] LIAN J, REN W W,LI L H, et al. PTP-STGCN:pedestrian trajectory prediction based on a spatio-temporal graph convolutional neural network[J]. Applied Intelligence,2023,53(3):2862-2878.
• [17] SINGH D,SRIVASTAVA R. Graph neural network with RNNs based trajectory prediction of dynamic agents for autonomous vehicle[J]. Applied Intelligence,2022,52(11):12801-12816.
• [18] STROHBECK J,BELAGIANNIS V,MüLLER J,et al. Multiple trajectory prediction with deep temporal and spatial convolutional neural networks[C]//2020 IEEE/RSJ International Conference on Intelligent Robots and Systems(IROS). New York, USA:IEEE,2020:1992-1998.
• [19]柳军杰，蔡英，范艳芳，等.基于GCN和Transformer时空信息融合的行人轨迹预测[J].北京信息科技大学学报（自然科学版），2024,39(6):1-8.LIU J J,CAI Y,FAN Y F,et al. Pedestrian trajectory prediction based on spatio-temporal information fusion using GCN and Transformer[J]. Journal of Beijing Information Science&Technology University(Science and Technology Edition),2024,39(6):1-8.(in Chinese)
• [20] WANG H Y,LI R N,XU Z L,et al. Federated learning for vehicle trajectory prediction:methodology and benchmark study[C]//2024 International Joint Conference on Neural Networks(IJCNN). New York, USA:IEEE,2024:1-8.
• [21] HAN M,XU K,MA S D,et al. Federated learning-based trajectory prediction model with privacy preserving for intelligent vehicle[J]. International Journal of Intelligent Systems,2022,37(12):10861-10879.
• [22] LI Y,XU X L,HUANG G J,et al. VSFL:trajectory prediction framework based on validity-aware semi-asynchronous federated learning in internet of vehicles[J]. Computer Communications,2024,224:106-117.
• [23] SHI J,SUN D X,GUO B C. Vehicle trajectory prediction based on graph convolutional networks in connected vehicle environment[J]. Applied Sciences,2023,13(24):13192.