复杂环境下嵌入式导航路径规划关键技术研究

发布时间：2018-03-13 03:34

本文选题：车辆导航系统　切入点：路径规划　出处：《解放军信息工程大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着GPS的发展,车辆导航系统得到了广泛的应用。作为车辆导航系统中的关键技术之一,路径规划已成为道路交通领域的研究热点。目前,基于二维路网拓扑的路径规划已在汽车卫星导航系统中广泛应用,然而随着北斗卫星导航系统的投入使用,用于军事用途的车辆和特种车辆不仅可能在无路的地方行进,而且行车过程中可能面临如天气、电磁、路障、态势信息及实时交通信息等影响路径规划的各种复杂因素,在如此复杂环境中有效地进行路径规划,成为当前该领域研究的难点之一。本文针对上述应用场景深入分析了基于路网拓扑和基于地形建模的两种不同环境下的路径规划相关问题,重点对路网数据模型、动态交通信息的权重分配、地形建模及路径搜索算法等问题进行了深入研究,完成的主要工作和学术成果主要包括:1.提出了一种面向动态导航的多尺度路网数据模型。根据不同任务对导航数据内容与尺度的需求和电子盘存储速度慢的特点,对用于显示的几何数据和用于路径规划的拓扑数据分别进行分层分块处理,存储在电子盘外部存储器中;在内存或外接数据库中存储时变的动态交通信息,并建立动态交通信息与静态路网关联关系。该模型在减少数据存储冗余的同时,支持大范围路径规划时依据道路等级特征分区读取不同层次路网数据,以有效减少路径计算的数据量,此外能够有效地表达动态交通信息,在数据准备为实现动态路径规划奠定了基础。2.研究了基于动态交通信息多元权重分配的路径规划算法。针对城市路网中动态交通信息对路径规划影响的问题,提出了一种基于模糊粗糙集的动态交通信息权重分配的路径规划方法。首先,建立动态规划模型,将行程时间和行车安全作为规划最优路径的标准;然后分别考虑影响行程时间和行车安全的因素,并将其分类量化,形成属性决策表,采用合适的隶属度函数计算各属性与规划原则之间的依赖度,确定各影响因素的权重;最后,利用分层A*路径规划算法对在路段上叠加了动态交通信息的城市路网进行最优路径搜索。仿真结果说明:该方法能够在保证算法实时性的前提下,更加合理的对融合交通信息的动态路网进行路径规划。3.研究了车辆在无路网覆盖区域进行路径规划的问题。由于特种车辆执行任务的特殊性,需要考虑在无道路网越野情况下基于实际地形的路径规划。在深入分析实际环境建模和无路网环境下路径规划算法的基础上,提出了基于高程-四叉树建模和改进蚁群算法的路径规划方法。该方法利用四叉树划分原则对野外实际地形基于高程差进行划分,在完整存储地形信息的基础上较大的降低了数据量,提高了存储效率,并降低了路径规划过程中的计算次数;基于已建立的模型进行了地形分析,为了使蚁群算法中启发函数更加准确有效,提出了综合考虑距离、坡度、粗糙度等信息的启发函数表示方法,并通过自适应调整挥发因子来优化信息素更新规则。实验结果表明,该方法相比以往算法,提高了路径规划的准确度,加快了收敛速度,并能够有效的避免算法早熟。4.综合运用上述研究成果,本章构建了嵌入式车载导航系统应用框架,并在Windows平台建立仿真开发环境,以某城市的导航路网数据为例,采用MapInfo构建多尺度路网模型,使用VC++编写路网数据格式的转换过程和路径规划算法,设计了导航数据格式转换软件,实现了多源路网数据、高程数据的格式转换;对基于复杂环境下的路径规划算法进行实现,并成功应用于WinCE平台的嵌入式车载导航系统,验证了系统的可用性。
[Abstract]:With the development of the GPS vehicle navigation system has been widely used. As one of the key technologies in the vehicle navigation system, path planning has become a hot research topic in the field of road traffic. At present, the network topology is two-dimensional path planning in car satellite navigation system based on the widely used, but with the Beidou satellite navigation system is put into use for, may not only in the road without a road where military vehicles and special vehicles, and the driving process may face such as weather, electromagnetic, roadblocks, various kinds of information and real-time traffic information of path planning of complex factors, effective path planning in such a complex environment, has become one of the difficulties of the present researches field. Aiming at the above scenarios in-depth analysis of the network topology and two different environment terrain modeling based on path planning based on Q The key problem of road network data model, the weight distribution of the dynamic traffic information are researched terrain modeling and path search algorithm, the main work and achievements include: 1. we propose a multi-scale network data for dynamic navigation model. According to the different tasks of navigation data content and scale of demand and the electronic disk storage speed characteristics, the geometric data for display and for topological data path planning are hierarchical block processing, stored in the electronic disk external memory; in memory or external number according to dynamic traffic information stored in a database changes, and the establishment of dynamic traffic information and static network association the model in reducing data redundancy and support large-scale path planning based on the characteristics of different road grade partition of road network data read, to effectively reduce the road The amount of data to calculate the diameter, also can effectively express the dynamic traffic information in the data preparation for the realization of dynamic path planning is the foundation of.2. research on the path planning algorithm of dynamic traffic information based on multiple weight distribution. In view of the influence of dynamic traffic information network in the city on the path planning problem, the paper proposed a path planning method for dynamic traffic the weight distribution information based on fuzzy rough sets. First, a dynamic programming model, the travel time and the traffic safety as the optimal path planning standard; then considering the factors influencing the travel time and the traffic safety, and classified quantitative form attribute decision table, using the appropriate membership function calculation of degree of dependence between the property and planning principles, determine the weight of each factor; finally, in section superimposed on the dynamic traffic using the hierarchical path planning algorithm A* City road network information search optimal path. The simulation results show that this method can guarantee the real-time performance of the algorithm under the more reasonable.3. for path planning of dynamic network traffic information fusion of vehicle in road network coverage area of the path planning problem. Because of the particularity of special vehicles to perform tasks, need to consider path planning based on the actual terrain in off-road conditions. Road network based on analyzing the actual environment modeling and path planning algorithm without network environment, put forward the elevation four fork tree modeling and improved ant colony algorithm based on path planning method. The method uses four tree division principle of field terrain division based on the elevation difference, based on the full storage of terrain information on greatly reduce the amount of data, improve storage efficiency, and reduce the path planning process Number of results; based on the established model of terrain analysis, in order to make the ant colony algorithm in heuristic function more accurately and effectively, is put forward considering the distance, slope, roughness representation heuristic information, and through the adaptive adjustment of evaporation factor to optimize the pheromone updating rule. The experimental results show that this method compared to the past the path planning algorithm, improve the accuracy and speed up the convergence rate, and can effectively avoid the premature convergence of.4. using the above research results, this chapter constructs the embedded vehicle navigation system application framework, and the establishment of simulation development environment in the Windows platform, the navigation road network data of a certain city as an example, the construction of multi-scale network model using MapInfo, prepared using the VC++ network data format conversion process and path planning algorithm, the design of navigation data format conversion software, to achieve multi way Network data, format conversion of elevation data, implementation of path planning algorithm based on complex environment, and successfully applied to embedded vehicle navigation system based on WinCE platform, validating the availability of the system.

【学位授予单位】：解放军信息工程大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U495;U463.67

【参考文献】