重庆理工大学学报2025，Vol.39Issue(17)：53-61,9.DOI:10.3969/j.issn.1674-8425(z).2025.09.007
基于交互风险场模型的智能车辆换道路径规划方法研究
Research on intelligent vehicle lane changing path planning method based on interactive risk field model
摘要
Abstract
To improve the decision-making of intelligent vehicles in changing lanes,this paper proposes intelligent vehicle lane-change path planning based on interactive risk field model.First,a vectorization map is built for intelligent vehicles' driving environment.Then,the obstacle vehicle feature coding network based on the feature pyramid and the lane line node feature coding network based on the void convolution are built respectively so as to complete the feature coding of the obstacle vehicles and the lane line nodes.According to the interactive relationship between different objects in the lane-change risk assessment,four interactive networks(vehicle-road,road-road,road-vehicle,and vehicle-vehicle)based on attention mechanism and graph convolution network,are built to accurately assess the dynamic and static risks and lane risks of obstacle vehicles.For the vectorized map,the driving risk field is modeled node by node,and the adaptive grid method is employed to discretize the driving risk field.Finally,the path cluster is assessed and the optimal path is selected by the cost function including driving risks.Through Argoverse dataset,a real traffic scenario simulation is conducted.Results show the proposed method effectively avoids potential risks,improves the intelligent vehicles' decision-making and risk assessment ability.关键词
路径规划/轨迹预测/自适应网格/图神经网络Key words
path planning/trajectory prediction/adaptive grid/graph neural network分类
交通工程引用本文复制引用
杨正才,李方祺,赵俊武,吴桐..基于交互风险场模型的智能车辆换道路径规划方法研究[J].重庆理工大学学报,2025,39(17):53-61,9.基金项目
湖北省技术创新计划项目(2024BAB086) （2024BAB086）
中央引导地方科技发展专项项目(2022BGE248) （2022BGE248）
湖北汽车工业学院博士科研启动基金项目(BK202215) （BK202215）
