设为首页
联系我们
加入收藏
您的位置:教学与研究 >>> 博士论文摘要 >>>李渊

 

>> 李 渊,2007,面向实际车道的三维道路网络模型及其应用研究


摘 要

随着城市向空中和地下的不断延伸,道路网络呈现出日益错综复杂的大立体特征,传统基于二维道路中心线的网络模型难以满足对实际交通设施和交通行为的详细表达与逼真模拟,使得城市微观尺度的三维道路设施管理与控制、智能交通、应急寻径与导航等应用存在巨大瓶颈问题。论文面向实际道路中的最小分析单元——车道带(Lane ribbon),研究面向实际车道的三维道路网络模型,支持道路网络的多层次表达(“道路中心线-行车道-实际车道”),车道几何、属性与连通关系的综合表达,交通事件
的准确定位与逼真表达等,在此基础上研究面向实际车道的三维动态分段和快速准确的多层次多模式寻径导航方法。主要研究内容包括:
1) 综合分析了道路网络模型的进展,重点比较了几种典型的道路网络模型,归纳总结了面向实际车道的三维道路网络模型的发展趋势以及亟待解决的关键科学问题。
2) 定义了车道带概念,采用最小拓扑分析单元--实际车道和最小几何分析单元--车道带构建“道路中心线-行车道-实际车道”一体化多层次的三维道路网络模型(3DLRNM),并利用面向对象的UML 建模语言建立多层次的拓扑与几何表达和不同抽象对象之间的逻辑关联,设计实现了3DLRNM 的数据库表达。
3) 提出了面向实际车道的三维动态分段(3DLDS)概念模型以及面向车道带的基准建立方法、基于“弧段-距离-车道序号”的线性参考方法、基于点序规则和基于语义规则的事件描述方法,使实际车道网络从现有线性参考方法中分离出来从而简化了动态分段的复杂性,并将交通事件的动态分段与准确的车道带位置灵活而紧密的关联,突破了现有动态分段方法在灵活精确定位方面的局限。
4) 扩展了面向实际车道的网络分析算法、分层寻径优化算法、地图视图到行为视图导航策略和三维多模式路径规划策略,解决了实际车道的拓扑构建和利用离散化方式解决实际车道拓扑更新问题,结合分层次认知规律提高网络分析效率和扩大导航信息的传输宽带,设计实现了建筑骨架数据的动态调度和建筑出口与实际车道连接点的语义自动匹配方法,提供了一套较为完整的寻径导航优化技术路线,提升了传统寻径导航在网络分析精确性不高、效率低下、导航信息传递不理想、应急反应适用性不强等问题,
5) 基于上述模型和算法,用VC++6.0 开发了原型系统,实现了多层次寻径和多模式导航功能,并对三维动态分段和基于语义信息的多模式路径规划分别进行了对比试验,分析了基于实际车道网络与多层次网络下的路径优化效率,实验结果证明了有关模型和算法的正确性与有效性。

 

关键词

GIS;GIS-T;三维城市模型;车道带;寻径与导航;三维道路网络模型;动态分段

Abstract

Modern road network systems are becoming multi-dimensional and dynamic as urban physical spaces are continuously expanding upward and downward. Consequently, the traditional 2D roadway centerline-based road network models gradually show their weaknesses in representing and modeling the real transportation features and the traffic behavior at very detailed level. There are huge bottlenecks between existing methods and transportation-related applications such as 3D road inventory management, intelligent transportation, emergency routing and navigation. Corresponding to the above problems, lane-oriented 3D road network model, shortly 3DLRNM is therefore developed in this research. 3DLRNM is characterized by: 1) uses real lane and lane ribbon as the basic topologic and geometric primitive; 2) supports multi-scale network representation (roadway centerline-carriageway-lane). Based on 3DLRNM, lane-oriented 3D dynamic segmentation and multi-level, multi-modal routing and navigation is further developed. The main contents of this research include:

1) Review of road network models. The research analyzes the typical road network models and concludes the development trends, followed by the summary of the key research problems of lane-oriented 3D road network models.

2) Illustration of 3DLRNM. The concept of “Lane ribbon” is defined, the multi-level (roadway centerline-carriageway-lane) database structure of 3DLRNM is designed with the use of UML modeling language after exploring the model framework and key objects involved in 3DLRNM.

3) Lane-oriented 3D dynamic segmentation. A conceptual model of lane-oriented 3D dynamic segmentation (3DLDS) is devloped, the lane-oriented datum-setting approaches, “link-distance-lane sequence” LRM, point sequence rule-based and semantic rule-based event description methods are introduced. 3DLDS simplifies the traditional dynamic segmentation approach by relaxing network topology from LRS and expands the event dynamic segmentation into 3D environment.

4) Lane-oriented 3D routing and navigation. The research analyzes the lane-oriented topology building and updating methods, develops multi-level routing algorithm and multi-modal navigation strategy by considering human cognition knowledge, and produces multi-modal emergency routing strategy by automatically connecting building entrance with lane via semantic information. A series of integrated technical workflows are discussed in order to improve the accuracy, efficiency, intelligence and integrity of lane-oriented 3D routing and navigation implementation.

5) Experiment and analysis. Based on 3DLRNM and related methods, the research develops prototype using VC++ 6.0 programming language. Two system functionalities are achieved: multi-level routing and multi-modal navigation. The 3D lane-oriented dynamic segmentation strategy, semantic-based multi-modal emergency routing strategy, and the comparative analysis between multi-level routing and single-level routing are also tested. The experimental results show the usefulness and effectiveness of the models proposed in this research.

Key words: GIS; GIS-T; 3D City Model; Lane Ribbon; Routing and Navigation; 3D Road
Network Model;Dynamic Segmentation.

徐胜华,2007,面向立体影像特征匹配的直线提取方法
版权所有:武汉大学虚拟现实实验室
地址:武汉市珞喻路129号测绘遥感信息工程国家重点实验室, 430079
电话:86-27-68778322 传真:+86-27-68778969