基于多智能体粒子群的快速碰撞检测算法研究  

Research on Fast Collision Detection Using Multi-Agent Particle Swarm Optimization

作　　者：付跃文[1] 梁加红[1] 李猛[1] 刘全平[1] 

机构地区：[1]国防科学技术大学信息系统与管理学院,长沙410072

出　　处：《系统仿真学报》2013年第8期1876-1880,共5页Journal of System Simulation

基　　金：国家自然科学基金(61170160)

摘　　要：在层次包围盒的基础上,提出一种基于多智能体粒子群的快速碰撞检测方法。算法首先利用层次包围盒方法快速减小物体间可能发生的碰撞检测区域,并基于随机碰撞检测核心思想将问题转变为物体特征对间距离机制的非线性优化问题,在建立的层次拓扑结构基础上,结合多智能体系统技术和粒子群算法的进化策略,设计了一种多智能体粒子群算法来求解碰撞检测问题。通过仿真测试表明,该算法具有很高的搜索效率和寻优性能,能够满足碰撞检测的实时性要求。A fast collision detection algorithm was proposed based on hierarchical bounding volumes and multi-agent particle swarm optimization（MPSO）.Firstly,the hierarchical bounding box method was used to reduce the potential areas of collision between objects quickly.And the problem was converted by stochastic collision detection method to a nonlinear optimization problem based on the distance of object characteristics.Combined with multi-agent system technology and the evolution mechanism of particle swarm optimization,a multi-agent particle swarm optimization based on the hierarchical topology structure was proposed to solve the problem of collision detection.The results of simulation prove that this algorithm has a higher searching efficiency and better optimal searching performance and it can meet the real-time requirements of collision detection.

关 键 词：碰撞检测 层次包围盒 粒子群算法 多智能体 拓扑结构 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]