论文部分内容阅读
We present an authoring framework for simulating realistic crowd behaviors in specific environment context at low computation cost.The proposed model aims to help users produce desirable crowd and ease their manual work.In our model, there are four important components, including triggers, limitators of range of motion, the cloud control terminal and groups.Triggers are mainly responsible for providing groups with relevant information about the environment context and therefore groups can act intelligently based on it.Limitators of range of motion represent the locations of static obstacles so that groups must avoid them when they move towards the goals.The cloud control terminal is responsible for computing optimal and congestion-free paths for groups according to the information relevant triggers provide.Groups, composed of many agents, can react to the specified goals as well as certain event appropriately by communicating with triggers and the cloud control terminal and furthermore, reflect the characteristics of the crowd.Users can easily construct an interesting and desired scenario with the crowd situated in it through integrating the above four components.The experimental results demonstrate that our model can not only well simulate diverse and interesting crowd behaviors in real time, but can also make users generate desired crowd in a simple and intuitive way without any technological knowledge about internal mechanism of our model.