基于图分类的智能车辆复杂场景风险等级评估与建模  

Risk Level Estimating and Modeling of Complex Scenarios for Intelligent Vehicles Based on Graph Classification

作　　者：吕超 孟相浩 崔格格 龚建伟[1] LÜChao;MENG Xianghao;CUI Gege;GONG Jianwei(School of Mechanical Engineering,Beijing Institute of Technology,Beijing 100081,China)

机构地区：[1]北京理工大学机械与车辆学院,北京100081

出　　处：《北京理工大学学报》2023年第7期726-733,共8页Transactions of Beijing Institute of Technology

基　　金：国家自然科学基金联合基金资助项目(U19A2083);国家青年自然科学基金资助项目(61703041)。

摘　　要：准确估计驾驶场景的风险等级是保障车辆安全驾驶的基础,也是车辆智能化的重要体现.针对多种交通参与者共存的复杂行驶场景,提出一种基于图分类的场景风险等级评估方法,完成对场景的建模和对当前场景风险等级的有效评估.实车实验表明,所采用的驾驶员操作特征数据可以很好地表示驾驶员对场景风险等级的理解,并且图表示模型可以对场景中多种动态交通参与者及其交互关系进行有效说明,所提出的方法能够较为准确地对复杂行驶场景的风险等级进行评估,促进智能车辆安全行驶系统在复杂环境下的发展.Accurate estimation of the risk level of driving scenarios is the basis to ensure the safe driving of vehicles,and it is also an important embodiment of vehicle intelligence.Aiming at the complex driving scene where multiple traffic participants coexist,in this paper,a scene risk level estimation method was proposed based on graph classification to complete the modeling of the scene and effectively evaluate the risk level of the current scene.The real vehicle experiment results show that the operation feature data of adopted driver can well represent the driver's understanding to the risk level of the scene,and the graphical representation model can effectively explain the various dynamic traffic participants and their interaction in the scene.The proposed method can more accurately evaluate the risk level of complex driving scenarios,and promote the development of intelligent vehicle safety driving systems in complex environments.

关 键 词：汽车工程 图表示模型 图核方法 聚类算法 风险等级评估 

分 类 号：U495[交通运输工程—交通运输规划与管理]