基于流体仿真的人群运动模拟基础算法研究
发布时间：2019-02-20 08:17
【摘要】：人群运动仿真技术和计算机技术的长足发展,大规模人群运动仿真也逐渐成为新兴的研究热点。如何更加全面而准确地研究并解决仿真中的问题,使仿真效果更具有真实感和高效性成为了研究者的重要课题。所以对大规模人群运动仿真的研究具有深远的学术意义和极大的现实价值。为了实现高效、真实的人群运动仿真,本文以大规模人群运动为研究对象,深入分析流体动力学,将流体仿真的方法应用到大规模人群运动的仿真中,实现两者的结合,提出了一种简洁灵活的人群运动控制方法。本文详细研究了基于径向基函数(Radial basis function,RBF)的群组运动仿真方法,研究其主要思想,完成实验,并在其基础上展开了本文的工作。本文采用了全局控制的框架,模拟在二维情景下的大规模人群行走。本文有别于RBF方法,通过求解纳维-斯托克斯(Navier-Stokes,NS)方程构建一个覆盖全局的速度场来驱动群组运动,在模拟过程中,允许用户通过交互操作改变NS方程组的参数来改变速度场,快速地改变人群的运动。另外,在局部(个体运动)控制方面,提出了基于光滑粒子流体动力学(Smoothed Particle Hydrodynamics,SPH)的碰撞处理策略。利用SPH方法来模拟个体的碰撞检测和碰撞避免等行为,避免了复杂包围盒碰撞检测技术,大大提高了模拟效率。在实验阶段,我们实现了本文算法,并将其和RBF仿真方法、Treuille势能场仿真方法、Narain仿真算法进行了比较。实验结果显示了本文基于GPU计算的算法在大规模人群运动仿真上有良好的效果,在复杂场景下将有较好的应用前景。本文算法使人群运动符合水流运动的规律,能更好的表现人群的行走效果。
[Abstract]:With the rapid development of crowd motion simulation technology and computer technology, mass crowd motion simulation has gradually become a new research hotspot. How to study and solve the problems in simulation more comprehensively and accurately has become an important task for researchers to make the simulation effect more realistic and efficient. Therefore, the research of large-scale crowd motion simulation has profound academic significance and great practical value. In order to realize efficient and real crowd movement simulation, this paper takes large-scale crowd motion as the research object, deeply analyzes fluid dynamics, applies fluid simulation method to large-scale crowd movement simulation, and realizes the combination of them. A simple and flexible method for crowd movement control is proposed. In this paper, the method of group motion simulation based on radial basis function (Radial basis function,RBF) is studied in detail, its main idea is studied, the experiment is completed, and the work of this paper is carried out on the basis of it. In this paper, a global control framework is used to simulate large-scale walking in two-dimensional scenarios. In this paper, different from RBF method, we construct a global velocity field to drive group motion by solving the Navier-Stokes,NS equation. It allows the user to change the velocity field by changing the parameters of the NS equations and to change the movement of the crowd quickly. In addition, in the aspect of local (individual motion) control, a collision processing strategy based on smooth particle hydrodynamics (Smoothed Particle Hydrodynamics,SPH) is proposed. The SPH method is used to simulate the behaviors of individual collision detection and collision avoidance. The complex bounding box collision detection technique is avoided and the simulation efficiency is greatly improved. In the experiment stage, we implement the algorithm and compare it with RBF simulation method, Treuille potential energy field simulation method and Narain simulation algorithm. The experimental results show that the algorithm based on GPU algorithm has good results in large-scale crowd motion simulation, and will have a good application prospect in complex scenarios. The algorithm of this paper makes the movement of crowd accord with the law of flow motion, and can better represent the effect of walking of the crowd.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.9
[Abstract]:With the rapid development of crowd motion simulation technology and computer technology, mass crowd motion simulation has gradually become a new research hotspot. How to study and solve the problems in simulation more comprehensively and accurately has become an important task for researchers to make the simulation effect more realistic and efficient. Therefore, the research of large-scale crowd motion simulation has profound academic significance and great practical value. In order to realize efficient and real crowd movement simulation, this paper takes large-scale crowd motion as the research object, deeply analyzes fluid dynamics, applies fluid simulation method to large-scale crowd movement simulation, and realizes the combination of them. A simple and flexible method for crowd movement control is proposed. In this paper, the method of group motion simulation based on radial basis function (Radial basis function,RBF) is studied in detail, its main idea is studied, the experiment is completed, and the work of this paper is carried out on the basis of it. In this paper, a global control framework is used to simulate large-scale walking in two-dimensional scenarios. In this paper, different from RBF method, we construct a global velocity field to drive group motion by solving the Navier-Stokes,NS equation. It allows the user to change the velocity field by changing the parameters of the NS equations and to change the movement of the crowd quickly. In addition, in the aspect of local (individual motion) control, a collision processing strategy based on smooth particle hydrodynamics (Smoothed Particle Hydrodynamics,SPH) is proposed. The SPH method is used to simulate the behaviors of individual collision detection and collision avoidance. The complex bounding box collision detection technique is avoided and the simulation efficiency is greatly improved. In the experiment stage, we implement the algorithm and compare it with RBF simulation method, Treuille potential energy field simulation method and Narain simulation algorithm. The experimental results show that the algorithm based on GPU algorithm has good results in large-scale crowd motion simulation, and will have a good application prospect in complex scenarios. The algorithm of this paper makes the movement of crowd accord with the law of flow motion, and can better represent the effect of walking of the crowd.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.9
【相似文献】
相关期刊论文 前10条
1 吴跃;马小虎;;一种基于GPU的阴影映射改进算法[J];微计算机信息;2009年36期
2 马伯宁;王晨昊;汤晓安;匡纲要;;基于GPU的二维离散小波变换快速计算[J];国防科技大学学报;2011年03期
3 ZW;;3D游戏利器 主流嵌入式处理器GPU逐个看[J];电脑迷;2011年19期
4 王志国;王贵锦;施陈博;苗权;林行刚;;积分图像的快速GPU计算[J];计算机应用研究;2011年10期
5 卢永菁;王东;;基于GPU的高速网络入侵检测系统设计[J];计算机工程与应用;2011年33期
6 储t熆，
本文编号：2426998
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2426998.html
