基于动态势场的车辆三维仿真方法研究

发布时间：2018-03-15 02:26

  本文选题：虚拟现实　切入点：动态势场　出处：《北京工业大学》2015年硕士论文　论文类型：学位论文

【摘要】：智能交通系统是未来交通发展的方向,仿真技术作为研究交通系统的有效手段,在智能交通领域有很好的应用前景。成功的搭建城市交通仿真模型,可以为交通道路设计提供依据。城市道路交通流是城市交通中的主要组成因素,而虚拟现实技术被证明非常适用于对城市道路交通流进行模拟研究,城市道路交通流虚拟模拟能够为行人交通安全教育以及城市道路交通流运动研究提供重要参考,具有重要的指导意义和广泛的研究前景。动态隐势场模型作为流体仿真的一个经典模型,将影响大规模群体运动的因素综合为一个连续的流场,大幅提升了仿真效果。该模型已经被国内外研究者运用到大规模人群仿真研究中,并取得了一定的研究进展。由于人群行为的自由度比较高,而车辆的行为可以看作是一个人群行为的一个简化,我们可以考虑将动态隐势场模型运用到车辆仿真上。因此,本文针对交通枢纽复杂的交通情况,研究基于动态隐势场的城市道路网络的车辆三维仿真方法。根据车辆在不同交通状况下的行为特点,本文为虚拟车辆赋予了多种智能化的运动策略,使车辆能够合理应对各种情况。在满足真实感和实时性的前提下,实现了虚拟车辆的智能化运动。具体完成的创新工作如下:(1)提出了一种基于动态势场的车辆运动模型。根据路况信息生成指导车辆运动的全局势场,完成局部区域的碰撞避免与车辆路径规划。并基于动态势场模型,实现了车辆的跟驰、换道行为。(2)提出了基于FIM(Fast Iterative Method)算法的车辆仿真优化方案。由于上述车辆运动仿真模型效率的主要标准为势场的构建速度,本文提出了基于FIM算法的势场并行计算方法,优化势场的构建效率。此外,本文结合CPU和GPU的运算特点,设计了不同的架构来实现车辆仿真系统,并最终得到了性能最为优良的基于多核CPU+GPU的并行异构的车辆仿真系统。(3)提出了基于并行环境的较为通用的地形信息描述方法。该方法建立了较为通用的场景表示方法和交通枢纽道路表示模型。本文以北京南站周围场景为例,同时运用上述车辆运动模型和仿真优化方案,实现了具有较高实时性的基于势场的车辆三维真实感仿真。实验结果表明,本文方法能够在大规模车辆的实时渲染场景中获得良好的仿真效果和仿真效率。在万车辆级别的场景想,仿真效率可以达到80fps。最并基于上述研究,设计并实现面向城市道路交通网络的车辆三维仿真系统。
[Abstract]:Intelligent Transportation system is the direction of traffic development in the future. As an effective means to study traffic system, simulation technology has a good application prospect in the field of intelligent transportation. Urban road traffic flow is the main component of urban traffic, and virtual reality technology has been proved to be very suitable for the simulation of urban road traffic flow. Virtual simulation of urban road traffic flow can provide important reference for pedestrian traffic safety education and urban road traffic flow movement research. As a classical model of fluid simulation, the factors affecting mass movement are synthesized into a continuous flow field. This model has been applied to the large-scale crowd simulation research, and has made some progress. Because of the high degree of freedom of crowd behavior, But the behavior of the vehicle can be regarded as a simplification of the behavior of a crowd. We can consider applying the dynamic hidden potential field model to the vehicle simulation. The 3D simulation method of urban road network based on dynamic hidden potential field is studied. According to the behavior characteristics of vehicles under different traffic conditions, this paper gives a variety of intelligent motion strategies for virtual vehicles. So that the vehicle can deal with all kinds of situations reasonably. Under the premise of satisfying the reality and real time, The intelligent motion of virtual vehicle is realized. The creative work is as follows: 1) A vehicle motion model based on dynamic situation field is proposed. According to the road condition information, the whole situation field is generated to guide the vehicle movement. The collision avoidance and vehicle path planning in local area are completed. Based on the dynamic situation field model, the car-following of vehicle is realized. This paper proposes a vehicle simulation optimization scheme based on FIM(Fast Iterative method algorithm. Because the main criterion of the efficiency of the vehicle motion simulation model is the speed of constructing potential field, a parallel computing method of potential field based on FIM algorithm is proposed in this paper. In addition, according to the operation characteristics of CPU and GPU, different architectures are designed to realize the vehicle simulation system. Finally, the most excellent parallel heterogeneous vehicle simulation system based on multi-core CPU GPU is obtained. A more general terrain information description method based on parallel environment is proposed. A more general scene table is established by this method. Demonstration method and traffic hub road representation model. This paper takes the scene around Beijing South Railway Station as an example, At the same time, the vehicle 3D realistic simulation based on potential field is realized with the above vehicle motion model and simulation optimization scheme. The experimental results show that, This method can obtain good simulation effect and simulation efficiency in real-time rendering scene of large-scale vehicles. In the scene of ten thousand vehicles level, the simulation efficiency can reach 80 fps. and based on the above research, A three-dimensional vehicle simulation system for urban road traffic network is designed and implemented.
【学位授予单位】：北京工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495;TP391.9
，

本文编号：1614031