融合交互预测信息的自动驾驶安全规划方法研究

doi:10.3901/JME.260122

机械工程学报 ›› 2026, Vol. 62 ›› Issue (4): 249-262.doi: 10.3901/JME.260122

• 运载工程 • 上一篇

扫码分享

融合交互预测信息的自动驾驶安全规划方法研究

唐小林^1,2, 张焜埸^1,2, 陈止戈^1,2, 杨剑英^1,2, 杨为^1,2

1. 重庆大学机械与运载工程学院重庆 400044;
2. 重庆大学高端装备机械传动全国重点实验室重庆 400044

收稿日期:2025-02-13 修回日期:2025-09-07 发布日期:2026-04-02
作者简介:唐小林,男,1984年出生,教授,博士研究生导师。主要研究方向为智能网联新能源汽车决策规划及节能控制。E-mail:tangxl0923@cqu.edu.cn
张焜埸,男,2001年出生,硕士研究生。主要研究方向为自动驾驶轨迹预测及运动规划。E-mail:zhangkunyi0523@163.com
杨为(通信作者),男,1973年出生,教授,博士研究生导师。主要研究方向为面向智能车辆的主动安全技术、先进电动汽车驱动与传动系统。E-mail:slmt053@cqu.edu.cn
基金资助:
国家自然科学基金(52388102，U2268210)和铁科院集团公司(2023YJ038)资助项目。

Research on Autonomous Driving Safety Planning Method Integrating Interactive Prediction Information

TANG Xiaolin^1,2, ZHANG Kunyi^1,2, CHEN Zhige^1,2, YANG Jianying^1,2, YANG Wei^1,2

1. College of Mechanical and Vehicle Engineering, Chongqing University, Chongqing 400044;
2. National Key Laboratory of Mechanical Transmission for High-end Equipment, Chongqing University, Chongqing 400044

Received:2025-02-13 Revised:2025-09-07 Published:2026-04-02

摘要/Abstract

摘要： 针对自动驾驶决策规划中动态环境交互与复杂行为预测的核心挑战，提出一种融合交互预测信息的自动驾驶安全规划方法。首先，构建图注意力网络(Graph attention network，GAT)提取智能体动态交互特征，并设计图卷积网络(Graph convolutional network，GCN)聚合车道拓扑交互特征。其次，联合前述交互特征以及自车规划特征，解码出自车运动影响下的周围车辆未来行驶轨迹。进一步，基于模型预测控制(Model predictive control，MPC)框架，通过预测时域维度对齐的方式融合预测信息，构建全局系统状态表征。同时设计目标函数以及约束条件保证车辆动作的最优化求解。实验环节基于真实驾驶轨迹数据集INTERACTION以及CARLA仿真平台进行验证。实验结果表明该方法在预测准确性、规划安全性以及通行效率等关键指标上均优于对比模型，尤其在交互密集型场景中展现出更强的意图推理能力。本研究为复杂动态交通场景下的自动驾驶安全规划提供了具有可解释性的解决方案。

关键词: 轨迹预测, 运动规划, 交互性决策, 模型预测控制

Abstract: This paper addresses the core challenges of dynamic environmental interaction and complex behavior prediction in autonomous driving decision planning by proposing a safety planning method that integrates interaction prediction information. First, a graph attention network(GAT) is constructed to extract dynamic interaction features of agents, while a graph convolutional network (GCN) is designed to aggregate lane topology interaction features. Subsequently, by combining these interaction features with the vehicle's own planning features, the future trajectories of surrounding vehicles under the influence of the vehicle's own motion are decoded. Furthermore, within a model predictive control(MPC) framework, the method integrates predictive information through time-domain alignment to construct a global system state representation. A objective function and constraints are designed to ensure optimal vehicle action solutions. Experiments validate the approach using the real-world driving trajectory dataset INTERACTION and the CARLA simulation platform. Experimental results demonstrate that this method outperforms comparison models in key metrics such as prediction accuracy, planning safety, and traffic efficiency, particularly exhibiting stronger intent inference capabilities in interaction-dense scenarios. This research provides an interpretable solution for safe autonomous driving planning in complex dynamic traffic environments.

Key words: trajectory prediction, motion planning, interactive decision making, model predictive control

中图分类号:

U491

唐小林, 张焜埸, 陈止戈, 杨剑英, 杨为. 融合交互预测信息的自动驾驶安全规划方法研究[J]. 机械工程学报, 2026, 62(4): 249-262.

TANG Xiaolin, ZHANG Kunyi, CHEN Zhige, YANG Jianying, YANG Wei. Research on Autonomous Driving Safety Planning Method Integrating Interactive Prediction Information[J]. Journal of Mechanical Engineering, 2026, 62(4): 249-262.

参考文献

[1] 高镇海，于桐，孙天骏. 考虑社会性行为的自动驾驶运动规划研究综述[J]. 机械工程学报，2024，60(10)：112-128. GAO Zhenhai，YU Tong，SUN Tianjun. A review of research on automatic driving motion planning considering social behavior[J]. Journal of Mechanical Engineering，2024，60(10)：112-128.
[2] HUANG Y，GU Y，YUAN K，et al. Human knowledge enhanced reinforcement learning for mandatory lane-change of autonomous vehicles in congested traffic[J]. IEEE Transactions on Intelligent Vehicles，2023，9(2)：3509-3519.
[3] SHENG Z，LIU L，XUE S，et al. A cooperation-aware lane change method for automated vehicles[J]. IEEE Transactions on Intelligent Transportation Systems，2022，24(3)：3236-3251.
[4] LIU Q，LI X，TANG Y，et al. Graph reinforcement learning-based decision-making technology for connected and autonomous vehicles：Framework，review，and future trends[J]. Sensors，2023，23(19)：8229.
[5] LIU H，HUANG Z，MO X，et al. Augmenting reinforcement learning with transformer-based scene representation learning for decision-making of autonomous driving[J]. IEEE Transactions on Intelligent Vehicles，2024，9(3)：4405-4421.
[6] Chowdhury J，Shivaraman V，Dangi S，et al. Deep attention driven reinforcement learning (DAD-RL) for autonomous decision-making in dynamic environment[C]//2024 IEEE International Conference on Systems，Man，and Cybernetics (SMC). IEEE，2024：3863-3868.
[7] Chen Y，Li S，Tang X，et al. Interaction-aware decision-making for autonomous vehicles[J]. IEEE Transactions on Transportation Electrification，2023，9(3)：4704-4715.
[8] 唐小林，杨剑英，杨凯，等. 面向无信号灯十字路口的自动驾驶多车协同控制方法研究[J]. 机械工程学报，2024，60(20)：217-228. TANG Xiaolin，YANG Jianying，YANG Kai，et al. Research on multi-vehicle cooperative control method of automatic driving for unsignalized intersection[J]. Journal of Mechanical Engineering，2024，60(20)：217-228.
[9] Fischer J，Bührle E，Kamran D，et al. Guiding belief space planning with learned models for interactive merging[C]//2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC). IEEE，2022：2542-2549.
[10] 杨凯，唐小林，钟桂川，等. 面向无信号灯十字路口场景的自动驾驶安全决策方法研究[J]. 机械工程学报，2024，60(10)：147-159. YANG Kai，TANG Xiaolin，ZHONG Guichuan，et al. Research on safe decision-making method of automatic driving for unsignalized intersection scenarios[J]. Journal of Mechanical Engineering，2024，60(10)：147-159.
[11] Rowe L，Ethier M，Dykhne E H，et al. Fjmp：Factorized joint multi-agent motion prediction over learned directed acyclic interaction graphs[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2023：13745-13755.
[12] Zhang L，Li P，Liu S，et al. Simpl：A simple and efficient multi-agent motion prediction baseline for autonomous driving[J]. IEEE Robotics and Automation Letters，2024，9(4)：3767-3774.
[13] Jia X，Wu P，Chen L，et al. Hdgt：Heterogeneous driving graph transformer for multi-agent trajectory prediction via scene encoding[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence，2023，45(11)：13860-13875.
[14] Song H，Luan D，Ding W，et al. Learning to predict vehicle trajectories with model-based planning[C]// Conference on Robot Learning. PMLR，2022：1035-1045.
[15] Li P，Pei X，Chen Z，et al. Human-like motion planning of autonomous vehicle based on probabilistic trajectory prediction[J]. Applied Soft Computing，2022，118：108499.
[16] Cheng L，Qin Y，Yang K，et al. Towards safe motion planning for autonomous driving in highway[J]. IEEE Transactions on Vehicular Technology，2025，74(2)：2491-2502.
[17] Gupta P，Isele D，Bae S. Toward scalable & efficient interaction-aware planning in autonomous vehicles using knowledge distillation[C]//2024 IEEE Intelligent Vehicles Symposium (IV). IEEE，2024：2735-2742.
[18] Hu Y，Yang J，Chen L，et al. Planning-oriented autonomous driving[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2023：17853-17862.
[19] Liang M，Yang B，Hu R，et al. Learning lane graph representations for motion forecasting[C]//Computer Vision-ECCV 2020：16th European Conference，Glasgow，UK，August 23-28，2020，Proceedings，Part II 16. Springer International Publishing，2020：541-556.
[20] Jiang B，Zhang Z，Lin D，et al. Semi-supervised learning with graph learning-convolutional networks[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2019：11313-11320.
[21] Schwarting W，Alonso-Mora J，Paull L，et al. Safe nonlinear trajectory generation for parallel autonomy with a dynamic vehicle model[J]. IEEE Transactions on Intelligent Transportation Systems，2017，19(9)：2994-3008.
[22] Mo X，Xing Y，Lv C. Recog：A deep learning framework with heterogeneous graph for interaction- aware trajectory prediction[J]. arXiv Preprint arXiv：2012.05032，2020.
[23] Zhao H，Gao J，Lan T，et al. Tnt：Target-driven trajectory prediction[C]//Conference on Robot Learning. PMLR，2021：895-904.
[24] Mo X，Huang Z，Xing Y，et al. Multi-agent trajectory prediction with heterogeneous edge-enhanced graph attention network[J]. IEEE Transactions on Intelligent Transportation Systems，2022，23(7)：9554-9567.
[25] Haarnoja T，Zhou A，Abbeel P，et al. Soft actor-critic：Off-policy maximum entropy deep reinforcement learning with a stochastic actor[C]// International Conference on Machine Learning. Pmlr，2018：1861-1870.
[26] Brito B，Agarwal A，Alonso-Mora J. Learning interaction-aware guidance policies for motion planning in dense traffic scenarios[J]. arXiv Preprint arXiv：2107.04538，2021.
[27] Liu H，Huang Z，Mo X，et al. Augmenting reinforcement learning with transformer-based scene representation learning for decision-making of autonomous driving[J]. IEEE Transactions on Intelligent Vehicles，2024，9(3)：4405-4421.

融合交互预测信息的自动驾驶安全规划方法研究

Research on Autonomous Driving Safety Planning Method Integrating Interactive Prediction Information

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	赵浩宇, 侯鹏帅, 陈星全, 邵宇乐, 董奇, 陈杰. 非结构化地形环境下移动机器人导航[J]. 机械工程学报, 2026, 62(3): 396-406.
[2]	庞文, 朱大奇, 林姝晗, 陈铭治. 基于多水下航行器编队控制的水下目标协同搬运研究[J]. 机械工程学报, 2025, 61(24): 363-376.
[3]	张利鹏, 刘一帆, 刘帅帅, 任长安, 高铭泽, 樊小健. 分布式驱动电动汽车回馈制动单侧电机失效的机电液复合控制[J]. 机械工程学报, 2025, 61(22): 250-261.
[4]	朱鹏兴, 胡建军, 李嘉佳, 彭航, 王鑫. 考虑驾驶风格不确定性的插电式混合动力汽车个性化生态自适应巡航协同优化方法[J]. 机械工程学报, 2025, 61(18): 214-229.
[5]	李元, 孙智, 訾斌, 陈兵. 基于改进高阶无迹卡尔曼滤波的膝踝康复矫形器预测控制与实验研究[J]. 机械工程学报, 2025, 61(11): 57-71.
[6]	焦恒超, 舒然, 曾兆枫, 唐小林, 舒红宇. 智能汽车稳定性MPC实时性改进的显式解方法[J]. 机械工程学报, 2025, 61(10): 263-275.
[7]	李鹏, 王刘银, 王刚, 马书根. 蛇形机器人全身避障策略设计与实验研究[J]. 机械工程学报, 2024, 60(21): 4-13.
[8]	张志勇, 王宇翔, 黄彩霞, 吴悠, 杜荣华. 融合灰色预测和卡尔曼滤波的车辆侧向碰撞预警[J]. 机械工程学报, 2024, 60(20): 240-250.
[9]	郭景华, 李文昌, 王班, 王靖瑶. 基于深度强化学习的网联混合动力汽车队列控制[J]. 机械工程学报, 2024, 60(2): 262-271.
[10]	杜国锋, 赵萌, 武建昫, 张东. 基于视觉的闭链多足机器人自主运动控制方法[J]. 机械工程学报, 2024, 60(19): 62-70.
[11]	姚建均, 张宜坤, 柯运, 钱琛, 王超. 基于模型预测控制的机器人姿态控制策略研究[J]. 机械工程学报, 2024, 60(19): 88-100.
[12]	关海杰, 王博洋, 龚建伟, 陈慧岩. 面向异构车辆的统一运动规划方法[J]. 机械工程学报, 2024, 60(18): 288-298.
[13]	王翔宇, 任帆, 刘冲, 许思昂, 王龙昕, 方勇纯, 于宁波, 韩建达. 面向NOTES手术的软镜操作机器人技术进展[J]. 机械工程学报, 2024, 60(17): 40-62.
[14]	李昊, 宾洋, 胡杰, 岳肖, 金庭安, 张凯. 无偏显式模型预测控制及其在质子交换膜燃料电池净输出功率跟踪中的应用[J]. 机械工程学报, 2024, 60(14): 282-297.
[15]	李玖法, 邹博文, 任玥. 考虑车-路交互作用的车辆轨迹预测算法研究[J]. 机械工程学报, 2024, 60(10): 76-85.