基于交通路网权重学习的城市机动车多样化轨迹生成  

Diversified trajectory generation of urban motor vehiclesbased on traffic road network weight learning

作　　者：王浩权 郑皎凌[1] 乔少杰 帅俨殊 刘双侨 曾宇 Wang Haoquan;Zheng Jiaoling;Qiao Shaojie;Shuai Yanshu;Liu Shuangqiao;Zeng Yu(School of Software Engineering,Chengdu University of Information Technology,Chengdu 610225,China;Sichuan Efang Intelligence Technology Co.,Ltd.,Chengdu 610000,China)

机构地区：[1]成都信息工程大学软件工程学院,成都610225 [2]四川易方智慧科技有限公司,成都610000

出　　处：《计算机应用研究》2025年第3期870-879,共10页Application Research of Computers

基　　金：四川省自然科学基金面上项目(2025ZNSFSC0488);香港中文大学(深圳)开放课题广东省大数据计算基础理论与方法重点实验室开放课题(B10120210117-OF02);云南省智能系统与计算重点实验室开放课题(ISC22Y02);四川省科技计划重点研发项目(2023YFG0027)。

摘　　要：基于GPS数据的轨迹生成方法因隐私保护与成本高的问题而难以应用,提出一种基于卡口数据生成车辆轨迹的方法。但其面临以下挑战:首先由于卡口覆盖率低导致拍摄的轨迹不连续,无法兼容现有模型,且未有工作研究如何有效填补缺失轨迹;其次现有模型忽略路网约束,生成轨迹无法进行仿真;最后现有模型无法生成多样化轨迹,导致可用性较差。为解决以上挑战,首先设计了TrajGAT-A模型,通过图神经网络构建包含实际交通信息的路网拓扑图,使用聚类算法构造出功能区网络并利用图注意力网络挖掘路网特征,生成路网权重图后执行A算法重构出连续轨迹。接着设计β-TrajVAE模型,通过聚类算法将路网划分为簇内外路段并执行分区采样,在损失函数中加入超参数以控制精度与散度之间的平衡,生成多个制导图后执行A搜索生成多样化轨迹。基于重庆数据进行实验验证,结果显示重构轨迹在precision、recall、F 1指标上均优于现有模型,生成轨迹在cross-entropy上优于现有生成模型,并通过仿真实验证实该方法可生成符合真实交通状况的轨迹。The acquisition of trajectory generation methods based on GPS data is challenging due to privacy protection and high costs.This study proposed a method for generating vehicle trajectories using checkpoint data,which confronted several challenges.Firstly,the low checkpoint coverage results in discontinuous trajectories that are not compatible with existing mo-dels,and no research has been conducted filling the gaps.Secondly,existing models overlook road network constraints,preventing trajectories suitable for simulation.Lastly,these models lack the capability to produce diverse trajectories,which diminishes their practicality.To tackle these challenges,it developed the TrajGAT-A model,which utilized a graph neural network to build a costmap rich with actual traffic information,employed a clustering algorithm to create a functional area network and leveraged a graph attention network to extract road network features.After constructing the costmap,it applied the A algorithm to reconstruct continuous trajectories.Following this,it designed theβ-TrajVAE model to segment the road network into intra and inter cluster sections via clustering algorithms and to conduct partitioned sampling.Hyperparameters were incorporated into the loss function to balance accuracy and divergence,leading to the generation of multiple costmaps for executing A search and producing diverse trajectories.Experimental validation using Chongqing data reveals that the reconstructed trajectories surpass existing models in terms of precision,recall,and F 1 metrics.The generated trajectories also outperform existing models in cross-entropy.Simulation experiments confirm that the proposed method is capable of generating trajectories that align with real traffic conditions.

关 键 词：轨迹生成 A算法 路径规划 VAE 

分 类 号：TP389.1[自动化与计算机技术—计算机系统结构]