【作者】 孙振平；

【导师】 贺汉根；

【作者基本信息】 国防科学技术大学 ， 控制科学与工程， 2004， 博士

【摘要】 论文从结构、设计与实践三个层次研究了自主驾驶汽车智能控制系统。 本文提出了一个四层递阶式智能驾驶控制系统结构。该结构基于任务逐层分解方法设计。在这个结构中，根据环境特性、时间跨度及空间范围等特点，将任务逐级分解为任务、子任务、行为、规划轨迹和动作五个层次，四层递阶式控制系统的四个层次从上到下依次完成这五个任务层次之间的映射。四层递阶式控制系统具有时间和空间上的多分辨率特点，从而使得控制系统的实时性和控制精度要求能够同时得到满足。 论文给出了自主驾驶控制系统总体性能评价指标，它们是：行驶的安全性、乘坐的舒适性和行车效率。论文进一步分析了如何利用总体性能评价指标来指导四层递阶式驾驶控制系统的分析与综合。另一方面，论文从环境感知系统的有关性能指标与驾驶控制系统性能的关系出发，讨论了驾驶控制系统的分析与综合问题。 针对双车道高速公路环境，论文研究了自主驾驶控制系统主要算法：第一、提出了基于行为的预期性能评价的行为决策策略；第二、构造了相应行为下的自主车行为规划算法；第三、分别设计了一个侧向跟踪控制器和一个速度跟踪控制器，提高了系统的抗干扰性和控制精度。 作为对本文研究成果的试验验证，论文对红旗自主驾驶汽车的开发与试验情况进行了介绍。更多还原

【Abstract】 This dissertation has researched on the intelligent control system for Autonomous Land Vehicle from three aspects: architecture, design and practice.A four-layer hierarchical intelligent control architecture is proposed for Autonomous Land Vehicle control in this dissertation. According to its environment characteristics, time span and space range, a task is hierarchically decomposed into five levels. These levels are task, sub-task, behavior, planned trajectory, and action. Four layers of the architecture are used to accomplish the mapping between these five levels. Profiting from its characterizes of multi-resolution on time and space, this control architecture can ensure system’s control period in real-time and control precision.Three overall performance indexes of Autonomous Land Vehicle control system are discussed, which are vehicle safety, passenger comfort and driving efficiency. The dissertation has discussed how to use these performance indexes to guide analysis and synthesis for control system. The dissertation has also discussed the analysis and synthesis problem of control system from the relationship between environments perception system and driving control system.Aiming at two-lane freeway, some main control system algorithms of Autonomous Land Vehicle are investigated: Firstly, an expected utility based behavior decision-making algorithm is put forward; Secondly, a behavior-planning algorithm is constructed; And thirdly, a path-tracking algorithm and a speed-tracking algorithm are designed, which have improved the disturbance rejection performance and control precision of system.As verification, The HONGQI Autonomous Land Vehicle and its experiment are introduced.更多还原