Study And Application Of Virtual Passenger Behaviors System In Metro Simulation Training

Crowd simulation is used to build crowd model and simulate crowd behaviors in virtual reality through artificial intelligence and computer graphics,it has been widely applied to public transport simulation,analysis of human flow for large public spaces,movies,video games and so on.Passenger intelligent behaviors simulation is based on crowd simulation technology,which can be used in simulating passenger behaviors including evacuation behavior.However,the present technologies of behaviors simulation are mainly concentrated on the evacuation behavior and use off-line,none-real-time calculation method,they couldn't be applied on metro simulation training system well.According to the requirements of reality,immersive visualization and real-time calculation,we raise an implementation of passenger behaviors simulation model for the scene of subway station to simulate passenger behaviors in a situation of several metros on the route with large,middle and small human flow.This model has been applied on metro simulation training system.At first,we study crowd perception model.The perception of individual is made up of vision,hearing and space detection.To represent the realistic perception effect,this paper build a perception model consist of the three.We process the three styles hierarchically to present the difference between agents.This model can embody the feature of human perception.Second,we study global static optimal path finding algorithm and dynamic obstacles avoidance strategy.We use navigation meshes to find the global static optimal path which can work in complicated environment.Then,this paper put forward an effective method of dynamic obstacles avoidance based on RVO algorithm with weight avoidance adding in to implement passing through or steering away from the passenger queue.Third,we study several kinds of crowd behavior models and analysis the difference between them.The model based on global method divide crowd into different level to simulate large-scale crowds,this model is widely used in movies and video games.However,the difference between agents can't be presented effectively.The model based on agent can endow individual unique attributes and calculate it singly.So this model has good effect on performing difference between agents.According to the system demand,this paper proposes a method of passenger behavior model based on behavior rules and characteristics to implement crowd simulation in different scale.Forth,to present the reality,we analysis and study passenger characteristics in different passenger flows and summary behavior law of passenger in subway station.In this paper,we design a calculation method of behavior choice based on random factors,properties individual and environment information.Finally,we study the model of emergency evacuation and simulate the situation according to the model of escape panic.To eliminate vibration and avoid model cross,we add stop rule into RVO algorithm.