群体行为规划技术的研究与实现

发布时间：2018-05-17 11:00

本文选题：群体行为仿真 + 行为引擎　；参考：《电子科技大学》2010年硕士论文

【摘要】： 群体动画在计算机游戏、影视动漫、城市建筑规划等很多方面都有着广泛的应用。然而,对大规模群体行为进行模拟尤其是实时模拟是一件复杂而艰巨的工作。因为对群体行为的仿真不仅涉及到高层的决策过程,还同时要考虑底层事物的表示以及事物之间的交互计算。更为麻烦的是,个体与个体之间以及个体与环境之间的复杂约束关系决定了在群体规模增大的时候,仿真运算复杂度将呈非线性增长。为此,本文提出构建一个高效的群体行为引擎来解决这一问题。本文首先讨论了行为系统的设计,然后着重研究了在路径规划算法方面进行改进及系统实现中使用加速技术来提高系统性能,最后介绍了群体的建模方法以及行为引擎如何被用来进行快速实现。构建行为引擎的目的在于简化群体仿真应用的开发过程。它的难点在于如何对行为仿真中的各种元素进行抽象、组织和融合。本文将行为引擎设计为信息管理模块、路径规划模块,以及系统控制模块三分部分。每个功能模块的作用和设计都做了比较详细的讲解。它预置了一些关键算法来提高行为仿真质量,同时也给用户提供了很好的扩展接口。路径规划是大规模群体仿真性能消耗的最主要部分。本文提出了一种基于势能场的实时高效的群体导航方法。该方法将局部势能场与全局势能场相结合,有效的解决了局部势能场中的极小值问题,同时又避免了全局势能场无法满足局部灵活性的问题。并行计算技术是目前最为流行的计算加速方法之一。本文分析了如何对碰撞避免和行为个体的更新进行适当修改,以适应并行执行。同时介绍了OpenMP并行方案如何被用在行为引擎中实现并行运算。另外,本文探讨了另一种简化运算的方法——LOD技术,并介绍了它如何用在行为仿真中以简化运算。最后,本文介绍了群体行为的建模方法,并以交通仿真和室外逃生系统为例,将重点放在如何利用行为引擎系统进行快速的群体建模实现,同时也作为行为引擎系统有效性的验证。随后,本文对目前的研究成果进行了总结,并对将来进一步的工作进行了展望。
[Abstract]:Group animation has a wide range of applications in computer games, film and television animation, urban building planning and many other aspects. However, the simulation of mass behavior, especially real-time simulation, is a complicated and arduous task. Because the simulation of group behavior involves not only the high-level decision-making process, but also the representation of the underlying things and the interaction between them. What is more troublesome is that the complexity of simulation will increase nonlinear when the population size increases because of the complex constraints between individuals and environment. Therefore, this paper proposes to construct an efficient group behavior engine to solve this problem. This paper first discusses the design of behavioral system, and then focuses on the improvement of path planning algorithm and the use of accelerated technology to improve system performance in system implementation. Finally, the paper introduces the method of group modeling and how the behavior engine can be implemented quickly. The purpose of building behavior engine is to simplify the development process of group simulation application. Its difficulty lies in how to abstract, organize and merge all kinds of elements in behavior simulation. In this paper, the behavior engine is designed into three parts: information management module, path planning module and system control module. The function and design of each function module are explained in detail. It presets some key algorithms to improve the quality of behavior simulation and provides a good extended interface for users. Path planning is the most important part of mass simulation performance consumption. In this paper, a real-time and efficient group navigation method based on potential energy field is proposed. By combining the local potential energy field with the global potential energy field, the method can effectively solve the minima problem in the local potential energy field and avoid the problem that the global potential energy field cannot satisfy the local flexibility. Parallel computing technology is one of the most popular computing acceleration methods. In this paper, we analyze how to modify the collision avoidance and the update of behavior individuals to adapt to parallel execution. At the same time, this paper introduces how OpenMP parallel scheme is used to implement parallel operation in behavior engine. In addition, this paper discusses another method of simplifying operation, LOD technology, and introduces how to use it in behavior simulation to simplify operation. Finally, this paper introduces the modeling method of group behavior, and taking traffic simulation and outdoor escape system as examples, it focuses on how to use the behavior engine system to realize group modeling quickly. It also serves as a validation of the effectiveness of the behavior engine system. Then, this paper summarizes the current research results, and prospects for further work in the future.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2010
【分类号】：TP391.9

【相似文献】