【作者】 冀杰；

【导师】 李以农；

【作者基本信息】 重庆大学 ， 车辆工程， 2010， 博士

【摘要】 智能车辆能够利用环境感知、信息融合、智能控制等关键技术实现自动驾驶功能,是智能交通系统(ITS)中的重要组成部分。智能车辆自动驾驶系统在提高道路通行能力、改善车辆主动安全性等方面具有巨大的应用潜力。针对智能车辆自动驾驶系统的运动控制及环境感知问题,论文建立了相对完整的整车动力学模型,设计了纵横向运动综合控制系统,并采用离线仿真试验和硬件在环技术对多种行驶工况下的自动驾驶控制效果进行了验证,另外,引入交互多模型算法对引导车辆进行了机动目标跟踪。论文主要包含以下工作内容:①对实际行驶过程中的车辆运动状态及受力情况进行分析,基于三维刚体动力学和运动学理论建立十自由度的车辆动力学模型;建立由发动机、液力变矩器、自动变速器及主减速器组成的车辆动力传动系统模型,综合考虑动力传动系统的非线性特性对车辆纵向运动控制的影响;考虑到智能车辆自动驾驶系统对轮胎模型力学特性的要求,建立可反映轮胎非线性力学特性的TMeasy轮胎模型,分析纵横向轮胎力之间的耦合关系。②为实现智能车辆的自动驾驶功能并提高其运动性能,论文基于模糊逻辑和滑模控制理论设计了具有上、下两层结构的纵横向运动综合控制系统,该控制系统能够对智能车辆的节气门开度、制动液压及前轮转向角进行协调控制,使智能车辆实现车道保持、车辆跟踪及车道变换等主要自动驾驶功能;与此同时,通过控制主动差速器的驱动力矩分配比例,产生所需的主动横摆力矩,提高智能车辆在自动驾驶过程中的操纵稳定性。③在车速变化、道路曲率变化及车道变换等多种行驶工况下,对纵横向运动综合控制系统的控制效果进行离线仿真试验,研究智能车辆的纵、横向运动跟踪性能及操纵稳定性。同时,对车道变换仿真试验中的虚拟理想轨迹进行设计,考虑各关键变量参数对其产生的影响;并对协调控制模块中的主动横摆力矩控制执行条件进行验证。④针对智能车辆自动驾驶过程中的车辆跟踪控制问题,论文采用交互多模型算法对具有多运动状态的引导车辆进行目标跟踪,从而为智能车辆提供准确、可靠的引导车辆运动状态信息。在仿真试验中,采用近似匀速和近似匀加速两种运动模型描述引导车辆的运动状态,同时,建立适用于道路车辆跟踪试验的传感器模型,并对机动目标模型的初始化问题进行研究。⑤为了验证智能车辆自动驾驶系统的有效性和鲁棒性,基于硬件在环技术和虚拟现实技术搭建同步视景自动驾驶试验平台。在该试验平台上,对多种行驶工况下的自动驾驶功能进行硬件在环仿真试验,并考虑人为驾驶操作对自动驾驶控制效果的影响。另外,通过实时调节车辆模型及控制系统的参数,分析智能车辆自动驾驶系统的鲁棒性。更多还原

【Abstract】 As a major component of Intelligent Transportation Systems (ITS), Intelligent Vehicle (IV) utilizes environmental perception, information fusion and intelligent control technologies to realize the automation of driving tasks. The intelligent vehicle autonomous driving system offers the great potential for significant enhancements in active safety and traffic capability.Aiming at the motion control and environmental perception problems of the IV’s autonomous driving system, a relatively complete vehicle dynamic model is developed and a combined longitudinal and lateral motion control system is designed in this dissertation. The performances of vehicular autonomous driving system are verified in different driving conditions by offline simulation and hardware-in-the-loop technology. In addition, the Interacting Multiple Model algorithm is introduced to track the leading vehicle on the road. The dissertation is organized as follows:①After analyzing vehicle’s translational and rotational behaviors in real driving conditions, a vehicle dynamic model with ten degree-of-freedom is developed based on three-dimensional rigid body dynamics and kinematics. A vehicular powertrain model that includes engine, torque converter, automatic transmission and final drive is developed, and the nonlinear characteristics of major components are analyzed simultaneously in this model. Considering the demand of tire forces for the intelligent vehicle autonomous driving system, this dissertation sets up a TMeasy tire model which can reflect the nonlinear force characteristics in saturation condition, and analyzes the coupling effects between longitudinal and lateral tire forces.②In order to realize the autonomous driving functions and improve stability of intelligent vehicle, fuzzy logic and sliding model control methods are used to design a combined longitudinal and lateral motion control system which has a hierarchical structure consisting of two layers. The system can control throttle angle, brake pressure and front wheel angle coordinately so as to make the intelligent vehicle achieve such autonomous driving functions as lane keeping, vehicle following, and lane changing. At the same time, it can improve vehicle’s stability during autonomous driving process by adjusting active differential’s distribution ratio of the driving torque so as to yield active yaw moment.③The performances of the combined longitudinal and lateral motion control system are examined in various driving conditions like velocity changing, road curvature changing and lane changing by off-line simulation, the stability of intelligent vehicle and motion errors in longitudinal and lateral directions are analyzed. At the same time, considering key factors such as lateral acceleration and lateral jerk in lane changing simulation experiment, a virtual desired trajectory is designed; this dissertation also validates the judgment basis of the active yaw moment control in coordinated control module.④As to the vehicle following control problem during autonomous driving process, this dissertation utilizes Interacting Multiple Model algorithm to track the leading vehicle with multiple states of motion. In this way, it can provide the Intelligent Vehicle with accurate and reliable information of leading vehicle. In simulation experiment, it uses uniform motion model with constant velocity and maneuvering motion model with acceleration to describe the motion states of the leading vehicle. In the meantime, a sensor model for such experiment is built and the initialization problem of maneuvering target model is studied in this dissertation.⑤In order to verify the effectiveness and robustness of intelligent vehicle autonomous driving system, a real-time visual autonomous driving experimental platform is developed, which is on the basis of hardware-in-the-loop and virtual reality technologies. On this platform, a series of hardware-in-the-loop simulation experiments are undertaken to test the autonomous driving functions in different driving conditions and the influence of human factors on vehicular autonomous driving system are considered. Besides, it analyzes the control system’s robustness by adjusting the parameters of the vehicle model and the control system.更多还原