Data-driven Scenario And Crowd Movement Modeling And Simulation

In recent years,with the progress of urbanization,the number of public places with dense crowds has been increasing,and public security has become the focus of public concern.How to organize crowd evacuation quickly and effectively during earthquake,fire and other emergency accidents,and maximize protect the life safety of citizens is an important research topic in the field of public safety.The application of computer simulation technology in this field can build model and simulate the process of crowd evacuation.It can help designers design reasonable structure of place,and help the safety management department to make emergency evacuation plans in advance.These applications have an important role to prevent the occurrence of disasters and minimize the casualty of the crowd after the occurrence of a disaster.Therefore,crowd evacuation simulation has important social significance,which has become one of the important research topics in the field of computer animation and virtual reality.In order to simulate the evacuation of the crowd in the scene,it is necessary to build the 3D models of the environment and compute the crowd motion.Although the current 3D modeling methods can efficiently generate large-scale 3D models,the process of extracting semantics is complex,and cannot be used for crowd evacuation efficiently.In the actual crowd motion,the crowd appears collectiveness phenomenon spontaneously,showing that the individual and other individuals around it maintain a consistent state of motion.The existing crowd evacuation simulation models do not take into account the collectiveness effect,and the simulation results differ greatly from the real crowd evacuation behavior.To solve the above problems,this paper firstly presents a fast simulation method for crowd evacuation based on geographic information.This method can quickly build 3D model,and extract semantic information from these models.Secondly,we propose a novel crowd simulation method for simulating collectiveness motion.The method can obtain more realistic simulation of crowd motion.Finally,this paper constructs a data-driven based crowd simulation system.The contributions of this paper are as follows:1.The traditional methods for crowd simulation are complex and inefficient both in 3D modeling and crowd motion computing which affects the efficiency of crowd evocation simulation.To address the above problem,this paper presents a fast simulation method for crowd evacuation based on geographic information.First,we design a 3D modeling method based on geographic information.The method can obtain the 3D scene modeling by abstracting the geographic coordinate of the road from 2D map and employing geometric transformation,surface construction and photo-realistic processing techniques.Then,we define the scene semantics based on geographic information and construct the topological graph to extract the scene semantics according to the obtained geographic information.The topological graph is used to path planning and individual navigation.Finally,we propose the Normal Distribution-based Reciprocal Velocity Obstacles method.The method combines normal distribution velocity with the Reciprocal Velocity Obstacles method and uses the scene semantics and global navigationmethod to compute crowd motion.2.In order to solve the problem of lacking of authenticity of crowd behavior simulation,a novel crowd simulation method for simulating collectiveness motion is proposed.First,we propose a collective model for crowd motion which defines the control parameter of collectiveness and collectiveness propagation algorithms to characterize the collective behaviors of pedestrians in crowded scenarios.Then,we couple the collective model with the Reciprocal Velocity Obstacles,and make the pedestrians collective by controlling the velocity of the crowd.Finally,the photo-realistic rendering method is employed to obtain the simulation results.3.A data-driven crowd simulation system is built.The system is implemented with Microsoft Visual Studio 2013,Microsoft Foundation Classes and Geographic Information System.It consists of three modules: three dimensional scene modeling,crowd evacuation simulation and crowd motion result display.The three dimensional scene modeling module is mainly used to extract geographic coordinate information data and construct the scene model.The crowd evacuation simulation module is used to setting up scene semantics and calculating the crowd motion.The crowd motion result display module is mainly used to render the crowd motion simulation effect.