Research And System Implementation Of Crowd Modeling And Simulation

Research on crowd modeling and simulation is an essential part of the crowd-related field, which aims to simulate the possible behavior of pedestrian in virtual environment by modeling the human crowd. With the growth of population and the increment of urbanization level in recent years, the research on crowd simulation plays an even more valuable role than before, that the loss caused by the problems of large scale of human crowd in public may be reduced in using the crowd simulation technique. There are lots of classic approaches purposed in foreign countries, and the relevant research works in China have been started.During the crowd simulation process, pedestrian individual would move at the maximal velocity, and the velocity would be held at this maximal velocity for majority of the simulation process. Such phenomenon is named after the "maximal velocity dilemma" and it exists in the traditional potential field as it is caused by the delay in velocity controlling of the traditional potential field method. In this paper, an improved potential field based crowd simulation method has been proposed. The velocity-density relationship model has been introduced into this method in order to solve the "maximal velocity dilemma" by establishing a refined the velocity controlling mechanism. Meanwhile, a better repulsive potential function has been given that a better group property could be presented, which brings out a more realistic simulated crowd. A hierarchical framework has been employed in this paper, in order to introduce the velocity-density relationship model. The simulation process has been divided into a global modeling process and a local simulation process. Global modeling includes environment modeling by a region separation and a global path planning by the derived topological presentation of the environment, while the local simulation includes the local modeling of potential field with the obtained global route and a velocity refinement by the velocity-density relationship model. The final simulation would be given by combine the result of global and local simulation.In addition, a prototype system has been built according to the proposed approach. Several experiments on the "maximal velocity dilemma" and crowd simulation have been constructed with the system. The result would show out the validation and solution of the dilemma and the effectiveness of the constructed system. According to the research work in this paper, the core should be the tackling of "maximal velocity dilemma" of potential field method and the group property representation. A more realistic crowd would be obtained by using the proposed improved potential field based model. However, the research on crowd simulation should not be limited to this. Since more social and psychological factors could be introduced, such as the sight range of a pedestrian, emotional controlling of the speed, the crowd simulation would be further extended. And these would be summarized in the discussion section.