一种基于动态感知的人群仿真局部避碰方法  

Dynamic perception based local collision avoidance method in crowd simulation

作　　者：史佳宾 刘箴[1] 肖苗苗[1] 刘婷婷[1] 刘翠娟[1] 刘邦权[1] 

机构地区：[1]宁波大学信息科学与工程学院,浙江宁波315211

出　　处：《计算机应用研究》2015年第10期3147-3150,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(61373068);浙江省自然科学基金资助项目(LY13F020037);宁波市科技计划基金资助项目(2013D10011;2014C50018);高等学校博士学科点专项科研基金资助项目(20133305110004);浙江省教育厅科研项目(Y201431792)

摘　　要：针对最优相互避碰方法在人群仿真中对虚拟人感知行为缺乏考虑,以及在人群密度较大时导致的拥塞行为,提出一种改进的局部避碰方法。首先,在已有虚拟人感知模型的基础上,加入虚拟人视觉随周围人群的密度而变化的规则,以及听觉感知对避碰行为的影响;其次,引入了有区别的避碰责任机制,使虚拟人在不同情况下采用不同的避碰行为;然后,加入一种拥塞响应规则缓解人群拥塞现象。最后通过多组实验与原方法进行对比,实验结果表明改进后的方法比原方法具有更真实的局部避碰行为,初步验证了该方法的可行性。The optimal reciprocal collision avoidance method had little consider with the perception of virtual people. And the crowd tended to fall in a state of congestion when the density became large. In order to solve this,this paper proposed a novel dynamic perception model for pedestrians. It presented a new rule for visual perception to simulate the dynamic change of the perception ability,and used auditory perception to control the collision avoidance behavior. By using a collision avoidance responsibility selection method,the pedestrians could make decision more like real people when avoiding the collision with others. In order to ease the congestion phenomenon,the model used a congestion detection and response rule. Simulation experiments show that the local collision avoidance behavior is more real and smooth than original method.

关 键 词：动态感知 局部避碰 人群仿真 相对速度障碍物 最优相互避碰 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]