Path Planning For Large Scale Crowds Simulation In Catastrophic Environment

Crowds simulation is the one of the most important branch in Computer Simulation. And it has been a research subject for a long time. Because of its multidisciplinary nature, it attracts more and more researchers from different research fields. Recent years, there has been many catastrophes and accidents. Some problems that we do not notice in usual may become hidden dangers. Therefore, we are eager to simulate catastrophe scene using crowds simulation, in order to expose these hidden dangers, and fix it in time.Owing to this requirement, we choose crowds simulation in catastrophe scene as our research subject. Based on the analysis of the key technology in crowds simulation,for specific requirement of our project, we proposed improvements in crowds modeling,path planning and collision avoidance. Our research can be divided into 4 parts:1. Studying the features of crowds in catastrophe, we proposed a new method for crowds modeling. The method is based on the Social Force Model and we take into account that people become sequacious in panic scene. We proposed a new velocity formula. Beyond that, we introduce Personality Model. Depending on a mapping from control parameters in crowds simulation to personality characteristics, we can simulate behaviors of agents with different personality.2. Studying large-scale path planning for crowds simulation, we proposed a kind of nonuniform grids. We choose the density of grid cells depending on different crowds densities, which enable the agent to sense the environment more accurately. Besides,individuals in the crowds usually do not have global knowledge, we propose the notion of partial visibility. Target points have been constrained to visible area.3. Studying collision avoidance technology in crowds simulation, we proposed a new method to compute velocity obstacle. And this method reduces the range of velocity obstacles and increases the range of available velocities, which reduces the infeasibility of linear programming and improves the possibility of individual collision avoidance. Beside that, we improve the random incremental linear programming, which make it feasible in the case of quadratic objective function.4. Based on the above theories, we completed a catastrophe crowds simulationprogram, which could present the features of catastrophe crowds realistically. It provides an ideal platform for other research on features of catastrophe crowds.