基于势能场的大规模人群运动模拟  

Large-scale crowd motion simulation based on potential energy field

作　　者：李丹[1] 袁凌[1] 胡迎松[1] 张训[1] 

机构地区：[1]华中科技大学计算机科学与技术学院,湖北武汉430074

出　　处：《华中科技大学学报（自然科学版）》2016年第6期117-122,共6页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(61502185)

摘　　要：针对大规模人群运动模拟中计算效率较低、实时性较差的问题,提出一种在智能体模型中引入势能场的新方法.该方法主要运用势能场来构建环境模型场力图,智能体通过感知获取环境信息,并规划初步行动路线;利用场力图,智能体再进行避免碰撞的行为决策.实验仿真表明:势能场的引入使感知计算在障碍物生成时就进行,避免了运动过程中还须要对静态障碍物进行计算,实现了预先计算的控制机制,使得大规模人群运动模拟的帧速率有较大提高,在保证准确性的同时还能提高效率.To solve the problem of low computation efficiency and poor real time performance in large- scale crowd motion simulation, a new method of introducing the potential energy field to the agent model was proposed. Potential energy field was mainly used to construct environmental model field. The agent body can retrieve the environmental information by perception, and plan the initial action course. With the field, the agent body can process the behavior decision to avoid collision. Experimental simulation shows that the introduction of potential energy field can make the perception computation be processed in the generation of obstacles, which can avoid the calculation of static obstacles in the action course. The implementation of precomputation control mechanism makes the frame rate of mass motion simulation greatly improved, which can improve the efficiency on the basis of ensuring the accuracy.

关 键 词：人群运动 势能场 智能体 感知 碰撞避免 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]