【作者】 谢峰；

【导师】 潘志庚；

【作者基本信息】 浙江大学 ， 计算机软件与理论， 2007， 硕士

【摘要】 计算机图形学和虚拟现实技术在近些年来取得了巨大进步，这为我们开发驾驶模拟训练系统提供了强有力的工具。本文描述了名为“面向驾驶训练的汽车驾驶仿真模拟机”(简称TODS)的驾驶模拟训练系统的实现。该系统旨在为了满足当今日趋严格的驾驶训练考核的需求而开发。TODS包括硬件系统和软件系统，TODS的硬件系统由个人计算机、显示系统和模仿真实驾驶室的驾驶舱组成；软件系统基于由香港理工大学开发的Lucid游戏引擎开发。TODS基于“人在回路中”模式开发。软件系统的开发分为软件架构、汽车建模与控制模块、控制信息输入模块、图形渲染、系统功能扩展模块、评判系统和环境建模模块七个部分。软件架构用于整合各个子系统并负责维护子系统之间的数据传递关系；汽车建模主要是汽车三维模型创建，并利用动态纹理映射技术实现汽车后视镜；汽车控制部分提供了虚拟汽车和物理控制设备之间的接口抽象，也负责获取来自硬件设备的控制信号；图形渲染系统以Lucid引擎为基础，利用多视角拼接技术支持120°以上的宽视角，使用户产生更接近真实车辆驾驶的沉浸感；系统功能扩展部分开发了性能分析模块、基于可编程图形芯片的图像处理框架；评判系统用于对用户操作过程进行评价，并反馈错误操作；环境建模包括考试场地建模和实际场地建模以及情节建模，考试场景模型按照《机动车驾驶员培训管理规定》创建，实际场地包括山区道路、城市场景和高速公路。其中利用模块化建模技术来简化复杂场景的建模工作，情节建模的主要工作则集中在自动汽车等agent的建模上。更多还原

【Abstract】 Recent advances in Computer Graphics and Virtual Reality technology have brought us the powerful tools for the development of driving simulators. This paper describes the implementation of a driving simulator named Training Oriented Driving Simulator(TODS). This system is designed to meet the needs of rigorous driving training and examination.TODS consists of hardware system and software system. The hardware system is constituted of PC, display system and driving cab. The software system is developed based on a game engine called Lucid developed by the Hong Kong Polytechnic University. TODS is designed under the human-in-loop real-time simulation mode. We divide software system into 7 parts, including software architecture, car modeling, I/O interface, graphics rendering, system extension, judging system and scene authoring.The software architecture, integrating the subsystems and managing information and data transfer among subsystems, is the key component of TODS. In car modeling, the paper describes the implementation of convex mirror using the dynamic texture mapping. In order to deal with different type of I/O device, we create an abstract layer between I/O devices and virtual car. The graphics rendering system is based on Lucid and provides 120° horizontal FOV and realistic visual feedback to give the driver a realistic feeling of immersion similar to the feeling in a real moving vehicle. We also develop some extension, like profile system and GPU based image processing framework. Judging system can help the user find out their errors. TODS’s scene model is designed and constructed according to the "Driving Test Specification" of China to meet the demands of rigorous driving training. And the modular construction method is used in the work of scene authoring in TODS.更多还原