[1]LI Minghan,XIAO Yang,XING Xianglei.Trajectory prediction methods based on analytical mechanics and graph neural networks[J].CAAI Transactions on Intelligent Systems,2025,20(6):1355-1365.[doi:10.11992/tis.202501020]
Copy

Trajectory prediction methods based on analytical mechanics and graph neural networks

CAAI Transactions on Intelligent Systems[ISSN 1673-4785/CN 23-1538/TP] Volume: 20 Number of periods: 2025 6 Page number: 1355-1365 Column: 学术论文—机器学习 Public date: 2025-11-05
Title:
Trajectory prediction methods based on analytical mechanics and graph neural networks
Author(s):
LI Minghan;  XIAO Yang;  XING Xianglei
College of Intelligent Systems Science and Engineering, Harbin Engineering University, Harbin 150001, China
Keywords:
trajectory prediction; graph neural networks; Lagrangian mechanics; interaction force; deep learning; multi-layer perceptron; kinematic model; analytical mechanics
CLC:
TP391.41
DOI:
10.11992/tis.202501020
Abstract:
Trajectory prediction seeks to forecast the future motion of intelligent agents by analyzing their past trajectories. While deep learning methods have been demonstrated to capture complex features, they frequently neglect physical constraints, thereby constraining interpretability. To address this, a trajectory prediction model is proposed that integrates analytical mechanics with graph neural networks (GNNs). The model combines GNNs, convolutional neural networks, and graph attention to extract spatiotemporal dynamics, infers interaction forces from Euclidean distance and relative motion, and incorporates Lagrange mechanics to enforce physical laws. Experiments on the Spring-balls dataset demonstrate the superior performance of the proposed model in comparison to traditional models, exhibiting a 14.29% accuracy gain in 10-frame short-term prediction for the 5-ball case and improvements of 6.25% and 4.81% in 50-frame long-term scenarios. In the domain of human motion prediction, our model demonstrates a reduction in mean position error (MPJPE) when compared to prevailing approaches for a wide range of actions. This finding signifies enhanced long-term accuracy and validates the efficacy of the model.
References:
[1] BARTH A, FRANKE U. Where will the oncoming vehicle be the next second? [C]//2008 IEEE Intelligent Vehicles Symposium. Eindhoven: IEEE, 2008: 1068-1073.
[2] KOOIJ J F P, SCHNEIDER N, GAVRILA D M. Analysis of pedestrian dynamics from a vehicle perspective[C]//2014 IEEE Intelligent Vehicles Symposium Proceedings. Dearborn: IEEE, 2014: 1445-1450.
[3] KARAMOUZAS I, SKINNER B, GUY S J. Universal power law governing pedestrian interactions[J]. Physical review letters, 2014, 113(23): 238701.
[4] BATZ T, WATSON K, BEYERER J. Recognition of dangerous situations within a cooperative group of vehicles[C]//2009 IEEE Intelligent Vehicles Symposium. Xi’an: IEEE, 2009: 907-912.
[5] PANG Bo, ZHAO Tianyang, XIE Xu, et al. Trajectory prediction with latent belief energy-based model[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville: IEEE, 2021: 11814-11824.
[6] LECUN Y, CHOPRA S, HADSELL R, et al. Energy-based models[M]//Predicting Structured Data. [S. l. ]: The MIT Press, 2007: 191-246.
[7] 姜立超, 尚晓兵, 金豹, 等. 基于遗传算法-v支持向量回归的船舶轨迹预测[J]. 哈尔滨工程大学学报, 2024, 45(10): 2001-2006
JIAN Lichao, SHANG Xiaobing, JIN Bao, et al. Ship trajectory prediction based on genetic algorithm-v support vector regression[J]. Journal of Harbin Engineering University, 2024, 45(10): 2001-2006
[8] LIU Xiaoli, YIN Jianqin, LIU Jin, et al. TrajectoryCNN: a new spatio-temporal feature learning network for human motion prediction[J]. IEEE transactions on circuits and systems for video technology, 2021, 31(6): 2133-2146.
[9] AKSAN E, KAUFMANN M, CAO Peng, et al. A spatio-temporal transformer for 3D human motion prediction[C]//2021 International Conference on 3D Vision. London: IEEE, 2021: 565-574.
[10] MAO Wei, LIU Miaomiao, SALZMANN M, et al. Learning trajectory dependencies for human motion prediction[C]//2019 IEEE/CVF International Conference on Computer Vision. Seoul: IEEE, 2019: 9489-9497.
[11] DANG Lingwei, NIE Yongwei, LONG Chengjiang, et al. MSR-GCN: multi-scale residual graph convolution networks for human motion prediction[C]//2021 IEEE/CVF International Conference on Computer Vision. Montreal: IEEE, 2021: 11447-11456.
[12] LI Maosen, CHEN Siheng, ZHANG Zijing, et al. Skeleton-parted graph scattering networks for 3D human motion prediction[C]//Computer Vision-ECCV 2022. Cham: Springer Nature Switzerland, 2022: 18-36.
[13] ZHOU Jincao, BAI Xin, FU Weiping, et al. Pedestrian intention estimation and trajectory prediction based on data and knowledge-driven method[J]. IET intelligent transport systems, 2024, 18(2): 315-331.
[14] 窦慧, 张凌茗, 韩峰, 等. 卷积神经网络的可解释性研究综述[J]. 软件学报, 2024, 35(1): 159-184.
DOU Hui, ZHANG Lingming, HAN Feng, et al. Survey on convolutional neural network interpretability[J]. Journal of software, 2024, 35(1): 159-184.
[15] 杨朋波, 桑基韬, 张彪, 等. 面向图像分类的深度模型可解释性研究综述[J]. 软件学报, 2023, 34(1): 230-254.
YANG Pengbo, SANG Jitao, ZHANG Biao, et al. Survey on interpretability of deep models for image classification[J]. Journal of software, 2023, 34(1): 230-254.
[16] 曹健, 陈怡梅, 李海生, 等. 基于图神经网络的行人轨迹预测研究综述[J]. 计算机工程与科学, 2023, 45(6): 1040-1053.
CAO Jian, CHEN Yimei, LI Haisheng, et al. A survey of pedestrian trajectory prediction based on graph neural network[J]. Computer engineering & science, 2023, 45(6): 1040-1053.
[17] 冯唐思捷, 梁伟. 基于物理信息神经网络的薄壁结构屈曲分析[J]. 力学学报, 2023, 55(11): 2539-2553.
FENG Tangsijie, LIANG Wei. The buckling analysis of thin-walled structures based on physicsinformed neural networks[J]. Chinese journal of theoretical and applied mechanics, 2023, 55(11): 2539-2553.
[18] 刘才山, 郭永新. 多体系统动力学与分析动力学专题序[J]. 力学学报, 2020, 52(4): 916.
LIU Caishan, GUO Yongxin. Special topics on dynamics of multi-body systems and analytical dynamics[J]. Chinese journal of theoretical and applied mechanics, 2020, 52(4): 916.
[19] LANDAU L D, LIFSHITZ E M. Course of theoretical physics[M]. British: Pergamon Press, 1980.
[20] BUTCHER J C. A history of Runge-Kutta methods[J]. Applied numerical mathematics, 1996, 20(3): 247-260.
[21] 张心祎, 谭耀, 邢向磊. 基于物理先验的深度特征融合水下图像复原[J]. 智能系统学报, 2023, 18(6): 1185-1196.
ZHANG Xinyi, TAN Yao, XING Xianglei. Deep feature fusion for underwater-image restoration based on physical priors[J]. CAAI transactions on intelligent systems, 2023, 18(6): 1185-1196.
[22] LUTTER M. Deep lagrangian networks: Using physics as model prior for deep learning[EB/OL]. (2019-07-10)[2024-09-19]. https://arxiv.org/abs/1907.04490.
[23] 刘建敏, 林晖, 汪晓丁. 基于图注意力机制的无地图场景轨迹预测方法[J]. 计算机工程, 2024, 50(7): 144-153.
LIU Jianmin, LIN Hui, WANG Xiaoding. Graph attention mechanism-based method for trajectory prediction in map-free scenes[J]. Computer engineering, 2024, 50(7): 144-153.
[24] 连静, 丁荣琪, 李琳辉, 等. 基于图模型和注意力机制的车辆轨迹预测方法[J]. 兵工学报, 2023, 44(7): 2162-2170.
LIAN Jing, DING Rongqi, LI Linhui, et al. Vehicle trajectory prediction method based on graph models and attention mechanism[J]. Acta armamentarii, 2023, 44(7): 2162-2170.
[25] 郝祎琛, 谢心喻, 丁家琦, 等. 瞬态多相流场图神经网络时空预测方法研究[J]. 哈尔滨工程大学学报, 2024, 45(9): 1761-1769
HAO Yichen, XIE Xinyu, DING Jiaqi, et al. Spatiotemporal prediction method for the transient multiphase flow field via graph neural network[J]. Journal of Harbin Engineering University, 2024, 45(9): 1761-1769
[26] 李浩淼, 张含笑, 邢向磊. 联合局部多尺度和全局上下文特征的步态识别[J]. 智能系统学报, 2024, 19(4): 853-862.
LI Haomiao, ZHANG Hanxiao, XING Xianglei. Gait recognition with united local multiscale and global context features[J]. CAAI transactions on intelligent systems, 2024, 19(4): 853-862.
[27] KIPF T, FETAYA E, WANG K C, et al. Neural relational inference for interacting systems[C]//International conference on machine learning. Stockholm: PMLR, 2018: 2688-2697.
[28] RIEDMILLER M. Multi layer perceptron[EB/OL]. [2024-10-11]. http://machine-learning-lab.com/_media/documents/teaching/ss12/ml/05_mlps.printer.pdf.
[29] HA S, JEONG H. Learning heterogeneous interaction strengths by trajectory prediction with graph neural network[EB/OL]. (2022-01-31)[2024-10-15]. https://arxiv.org/abs/2105.14491.
[30] HUANG W, HAN J, RONG Y, et al. Equivariant graph mechanics networks with constraints[EB/OL]. (2022-03-12)[2024-12-15]. https://arxiv.org/abs/2208.13179.
[31] IONESCU C, PAPAVA D, OLARU V, et al. Human3.6M: large scale datasets and predictive methods for 3D human sensing in natural environments[J]. IEEE transactions on pattern analysis and machine intelligence, 2014, 36(7): 1325-1339.
[32] BRODY S, ALON U, YAHAV E. How attentive are graph attention networks? [EB/OL]. (2021-05-30)[2024-12-15]. https://arxiv.org/abs/2105.14491.
[33] MA Tiezheng, NIE Yongwei, LONG Chengjiang, et al. Progressively generating better initial guesses towards next stages for high-quality human motion prediction[C]//2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans: IEEE, 2022: 6427-6436.
Similar References:

Memo

-

Last Update: 1900-01-01

Copyright © CAAI Transactions on Intelligent Systems