摘要
为仿真现实生活中由纵横交错的道路路网、大规模的车辆及其运动组成的交通场景,文中提出一种复杂路网内大规模车辆运动的仿真技术.首先设计一种层次路网语义模型,根据输入的路网车道线矢量数据,通过语义建模,从几何、拓扑等不同层次描述错综复杂的交通路网.其次,由于现有的车辆运动仿真方法中基于个体的微观方法针对大规模车辆运动计算效率较低,而基于流的宏观方法无法仿真复杂的交通现象,提出了一组新的宏观流方程来描述路网内车辆的运动,使得宏观流模型能够与车辆换道行为模型有机地进行结合,从而逼真地描述各种复杂的交通现象.实验结果表明,文中方法在实现复杂路网内车辆运动高细节模拟的前提下,其计算效率与一般宏观流模型保持在同一数量级.通过在实际应用中对真实交通路网进行虚拟现实建模,并在其上进行大规模车辆运动仿真,进一步验证了方法的有效性.
Traffic is one of the most important components in our daily life.In the real world,traffic scenes are complex including crossed road network and mass vehicles as well as their shifty motion.Simulating vivid complex traffic scenes is important for many traffic management applications,such as road network optimizing,traffic impact analysis and evacuation planning in some emergent situation.It is also very important for some virtual reality applications,like games,virtual tourism and virtual city.In this paper,a novel simulating method for complex networks with large scale traffic is proposed.Firstly,a hierarchical semantic model is introduced to describe crossed road networks.It includes a geometrical layer which couples3D geometrical lane data with its mileage representations together,so that the position query in traffic simulation could be fastened.It also has topological layers to make mass lane slices become an organic network.The topological layers are semantically completed,including links,road segments,their connectors and intersections,so that they could be built from the geometrical layer automatically.Secondly,a novel hybrid traffic model is proposed.It develops the traditional single lane traffic flow equations to describe the flow interaction between neighboring lanes and make it possible to combine a lane changing model inside.Unlike existing macroscopic models which could not describe complex traffic phenomenon,and unlike existing agent based models which have low efficiency in handling large scale traffic,the proposed hybrid model could naturally combined lane change behavior inside to simulate detailed complex traffic phenomenon while keeps high efficiency as most existing macroscopic models.The solving method of the hybrid traffic model is also introduced.Experimental results show that,compared with other macroscopic ones,the efficiency of our method has the same order of magnitude.It could simulate traffic on a600km network in real time.However,more than any other macroscopic models,it could dem
作者
毛天露
王华
康星辰
徐明亮
王兆其
MAO Tian-Lu;WANG Hua;)KANG Xing-Chen;XU Ming-Liang;WANG Zhao-Qi(Beijing Key Laboratory of Mobile Computing and the New Terminal, Institute of Computing Technology, Chinese Academy of Sciences, Beijing100190;College of Computer and Communication Engineering, Zhengzhou Unversity of Light Industry, Zhengzhou 450002;College of Information Engineering, Zhengzhou University, Zhengzhou 450001)
出处
《计算机学报》
EI
CSCD
北大核心
2017年第11期2466-2477,共12页
Chinese Journal of Computers
基金
国家"八六三"高技术研究发展计划项目基金(2013AA013902)
公益行业科研专项课题(2013467058)
国家自然科学基金(61472370)
国家自然科学基金青年基金(61402269)资助
关键词
大规模车辆
复杂路网
层次语义模型
车辆换道行为
宏观流模型
虚拟现实
large scale traffic
complex road networks
hierarchical semantic model
lane changes
macroscopic traffic model
virtual reality
