Modeling And Simulation Of Large Scale Crowd Behavior Of Ancient Soldiers In A Virtual Ancient City

In recent years, with the continuous development in the fields of computer animatedfilms, online games and virtual reality applications, developing computer animationtechnologies for generating realistic virtual character’s motion behaviors has been an activearea of research with many challenges. At the same time, simulations of large crowdbehaviors in complex virtual environments have been a focus for researchers in order togenerate the most beautiful natural human behaviors for the most common scene that we haveseen in our everyday life, for example, crowd behaviors and motions in interactive virtualreality application for virtual training, safety and security rehearsal etc.. This project isaimed at developing virtual reality techniques for the reproduction of crowd simulationswithin ancient historical cities and environments, more specifically groups of virtual soldiersin order to recreate and preserve virtual historical scenes., The research work of this projectwill investigate the techniques used in digital games,3D character animation techniques andmotion capture technology in modeling virtual characters’ motions.This project addresses issues concerned with characters motion animation. Althoughmotion capture technology as a major technology for generating realistic motion behaviors of3D character animation with high fidelity and rich details, and it has become the standardmethod in computer games, training simulations, movie special effects in movies, motioncapture techniques also has some disadvantages, such as expensive capture equipment, largedatasets and high memory storages, especially data reusability must be recomputed andsynthesized for new motions.In this project, an existing motion capture dataset is used to generate massive ancientsoldier groups for character animation and motion control, including the rendering of virtualhuman and clothing, the generation of hierarchical virtual characters skeleton data structureand virtual characters’ skin binding techniques. We are also studying the methods foranimating the hierarchical skeleton structure and motion path planning techniques. In order tomodel and simulate large crowd behaviors in ancient cities a particle swarm system isdeveloped in this project to achieve the animation of groups of soldiers. The main results and contributions of this project are as the follows:1) The use of an existing motion capture database for virtual characters’ motion analysis,the ASF-AMC data format has been employed. The motion data is used in the Maya softwareenvironment with a script language (MEL) programming for re-editing and generating therequired motion data animation dataset for virtual characters in this project.2) In order to create realistic ancient solder characters, a3D human skeleton model isdeveloped with a character’s skin binding model, The movement of a single ancient soldier isfirstly created, and then the MEL script programming is used for motion path planning. Theexperiments show that this method can realistically simulate the movement of the humanmotions.3) For crowd motion behavior animation, an alternative method of using particle swarmsystem has been developed to achieve the realistic movement of groups. The experimentsshow that the method can quickly generate a group of animated motions and actionreunification, which are neat and suitable for large-scale crowd behaviors. The techniquesdeveloped in this project are also applicable to other virtual reality applications such as fortheatrical performances, military and safety trainings.4) As simulations of large crowd motion behaviors are computationally expensive and itis important to achieve real-time simulations for interactive virtual environments. In thisproject, we have developed a method to use the billboard technology for virtual charactersthat are in a far distance from the user to reduce the computational costs, achieve fastsimulation performance, yet still high quality animations.The proposed method is successful in generating three-dimensional scenes, which has agroup of ancient soldiers marching, running and practicing actions. The results show that themethodologies developed in this project can create efficiency and high fidelity groupanimations.