Study On Crowd Simulation Technology Based On Virtual Reality

In recent years,with the rapid development of virtual reality technology,crowd simulation in virtual environment has gradually attracted the attention of universities and research institutes.Virtual scene has three characteristics of immersion,interaction and imagination,which determine that the virtual environment needs high authenticity.However,in the virtual scene,the complex terrain,a large number of vegetation and buildings make the number of rendering amazing,rendering acceleration has become an important technology.In realistic virtual environment for crowd simulation technology has been applied to various fields,such as a large scene crowd evacuation simulation,military drills,public places of people flow forecast and analysis,etc,through the analysis of simulation results,can effectively improve the evacuation route,against strategy and design scheme of public places,so the study on the crowd simulation in virtual environment has important practical significance.In this paper,the Unmanned Aerial Vehicle is firstly used for image acquisition and processing.3d modeling is carried out in Maya based on the image,and texture mapping and material production are carried out in Mari.Second using a good model for 3 d reconstruction in the Unreal Engine,to generate stylized of vast amounts of vegetation,the game thread and rendering thread synchronization parallel computing,partitioned loading of large-scale terrain,model of a large number of secondary cutting algorithm,the first cut for the first time,using the improved Fade In-and-Out Levels of Detail algorithm is outside of the view model,reuse the Fade Culling algorithm,will but obscured vision range model is secondary.Then crowd simulation is carried out in the optimized virtual environment,and crowd simulation technology is divided into three parts: collision detection,global path planning and local path planning.Collision detection uses Capsule technology.A* algorithm based on 24 neighborhoods is used for global path planning to calculate the optimal path.In the process of virtual crowd movement,the optimal speed algorithm is used to calculate the undetected collision speed and the RVO2 algorithm is used to calculate the detected collision speed.Finally,the crowd simulation algorithm in this paper is compared and analyzed by experiments with the original crowd simulation algorithm of unreal engine.Through the comparison and analysis of the experimental results,it is proved that the rendering speed of the above method is improved by forty percent compared with the rendering speed of serial threads.The "jump" phenomenon of traditional Levels of Detail algorithm and elimination algorithm was solved by using the Double-layer cut algorithm,and the rendering speed was greatly improved,its frame rate is also improved by fifty-five percent.Use Capsule in the crowd simulation bounding box collision detection,detection accuracy reached 91%,using 24 neighborhood in global path planning of the A * algorithm,the relativesmooth path,used in the local path planning algorithm and the optimal speed RVO2 algorithms of the simulation time own GPU computing simulation method than the engine of a frame of time increased by 50%,and the frame rate also has the improvement.