【作者】 张凯；

【导师】 吴清；

【作者基本信息】 河北工业大学 ， 计算机应用技术， 2011， 硕士

【摘要】 随着虚拟现实技术的发展,针对虚拟人技术的研究逐渐增多。本文通过分析虚拟人的关键技术,重点研究了感知模型和路径规划的方法。在分析探讨已有感知模型和路径规划算法的基础上,建立适合的视觉感知模型,融合相关路径规划算法,并应用于虚拟场景中虚拟人的路径规划。首先,建立虚拟人几何模型。本文采用骨骼蒙皮的方法,在3ds max中将Biped骨骼嵌入人体模型,实现人体模型的行走运动,为实验中模拟虚拟人在虚拟场景中进行路规划提供基础。其次,主要研究感知模型。感知环境信息属于虚拟人行为控制技术层次。本文通过对典型感知模型进行比较,分析人类的视觉感知生理特点和注意机制,采用人眼视野范围和场景信息空间查询相结合建立视觉感知模型。该模型虽然没有涉及视觉记忆和其他的感知方式,逼真性差了些,但处理效率比较高。最后,在建立视觉感知模型基础上,着重研究虚拟人的路径规划技术。其中环境信息表示是首要问题,本文采用栅格法和四叉树相结合来表示虚拟环境。全局静态环境下运用启发式的搜索算法—A*算法,出现动态障碍物的环境下运用人工势场法。A*算法的启发函数是其关键所在,本文在启发函数中加入栅格节点的朝向角度,对其和欧氏距离添加不同的权值,既加入了最短距离因素,也加入方向因素,使得虚拟人路径规划能够增加多样性。对于人工势场法运用已有的思想,加入虚拟障碍物打破振荡状态,可以实现在动态环境下的躲避障碍物,对多运动物体采用大序号物体优先运动。并将这两种径规划算法通过实验仿真,在虚拟场景中进行虚拟人的路径规划,证明了算法的有效性。更多还原

【Abstract】 Along with the development of virtual reality technology, research on the technology of virtual human is increasing. Through the analysis of key technology of virtual human perception, this thesis focuses on the method of perception model and path planning. With discussion about the existing perception model and path planning algorithm, develop suitable visual perception model, combined with path planning algorithm related. And it is applied to virtual human path planning in virtual scene.Firstly, develop virtual human geometrical model. The thesis adopts skeleton skinned method, which makes Biped-bone embedded in human body model in 3ds max, and realizes the model of human body’s walking movement and provides basis for experiments simulating.Secondly, this thesis mainly studies perception model, and makes a comparison of typical perception models, analyses human visual perception physiology characteristic and attention mechanism, establishes a visual perception model through human visual range, combined with the query of scene information space. The model does not refer to visual memory and other perceptive manners and has a not very good verisimilitude, but it is with high efficiency.Finally, based on the visual perception model, the thesis focuses on path planning technology. The representation of the environment information is the key problem; this thesis adopts grid method and quad tree to represent the virtual environment. This paper uses heuristic search algorithm--A* algorithm in global static environment and artificial potential field method in dynamic obstacles environment. Heuristic Function is the key of A* algorithm.The orientation angle of grid node is added to Heuristic Function, and different weights are added to it and Euclidean distance. So distance and direction factors are considered to increase the diversities of virtual human motions. Using existing thoughts and adding virtual obstacles in artificial potential field method, virtual human can realize evading obstacles in dynamic environment. For multiple objects in motion, the movement is based upon priority of large number. These two path planning algorithms are simulated through experiments, which prove the validity of proposed algorithm.更多还原