基于视觉和雷达的智能车辆自主换道决策机制与控制研究

发布时间：2018-03-25 14:43

本文选题：智能车辆　切入点：自主换道　出处：《中国人民解放军军事医学科学院》2014年博士论文

【摘要】：智能车辆从根本上改变了传统的车辆驾驶方式,将驾驶员从“驾驶员一车辆-道路”的闭环系统中解放出来,利用先进的电子与信息技术控制车辆行驶,让驾驶活动中常规的、持久且易疲劳的操作自动完成,能够极大地提高交通系统的效率和人员的安全性。研究智能车辆的自主换道关键技术,最终能够实现多车的自主交互协同,提高部队人员、装备的使用效率和战场环境的适应能力；同时,通过准确的环境信息感知,加上科学、合理的决策分析与稳定可靠的控制算法,使车辆自主换道的安全性比充满不确定因素的驾驶员换道更具优越性,有效地控制人为因素造成的交通事故。本文通过分析驾驶员的驾驶行为过程,研究了驾驶员换道意图的产生及阶段,分析了影响驾驶员换道的因素,进而深入研究了驾驶员换道的决策机制,并针对智能车辆的结构特点,模拟驾驶员换道的决策过程,解析了智能车辆自主换道的决策机制。通过分析人体器官在驾驶员驾驶车辆过程中的功能,建立了“驾驶员-车辆-道路”的系统模型,从控制论的角度分别研究了驾驶员、车辆和道路在系统中的功能与作用,并提出了系统控制的评价指标；通过试验数据,得出了诱发车辆换道的主要原因是本车道前方有慢车,而驾驶员对时间与空间的追求是影响车辆换道的主要因素；本文将车辆的换道过程分为换道意图的产生、换道时机的决策、换道轨迹的规划和换道轨迹的跟踪控制四个阶段；建立了“驾驶员-车辆-道路”系统的高速公路典型场景,得出了驾驶员的换道过程是以驾驶员行为特性为主导的信息感知、决策与操控的三个模块相互作用的行为决策与控制过程；分析了驾驶员换道决策阶段的表征参数,选取了车道线信息、车道边缘信息、交互车辆信息等作为智能车辆自主换道研究的特征参数；根据智能车辆的结构特点,模拟驾驶员换道决策过程,建立了“机器-车辆-道路”的系统模型,解析了智能车辆自主换道的决策机制。通过建立路权雷达图进行信息融合,对换道过程中换道意图的产生、换道时机的决策、换道轨迹的规划三个阶段进行分析,建立了智能车辆自主换道的决策模型,并针对换道过程出现的突发异常情况,建立了静、动态障碍车辆的避障模型。模拟人类认知行为的注意力分配机制,根据路权的概念,建立了变粒度路权雷达图,使用较少的存储空间和计算资源完成对人类认知行为的模拟和计算,通过路权雷达图实现了信息融合、仿真分析和路径规划等功能；定义了智能车辆的最小行车安全距离,在此基础上确定了智能车辆产生换道意图的期望值。通过选取2名熟练驾驶员、选择典型高速公路路线、制定试验控制条件等设计试验方案,采集影响驾驶员换道的特征参数311组,其中,车道保持数据97组,换道行驶数据214组；采用v-支持向量机进行训练,选取(δ,v)=(0.12,0.03)作为v-支持向量机的模型参数,判断换道行为的准确率为91.05%；从换道时间、换道横向加速度、曲率突变等方面分析比较了智能车辆常用的换道轨迹规划方法,根据本文研究的对象,确定了梯形加速度换道轨迹的方法；针对换道过程的突发意外事件,建立了静、动态障碍车辆的避障模型；为了满足实时性要求,在环境建模中设计了变尺度栅格图,当车辆高速行驶时,变尺度栅格图比传统栅格的CPU占用降低约34%；将变尺度栅格图与路权雷达图进行融合,通过静态选择式避障模块和动态障碍避障策略,实现对突发意外障碍车辆的躲避。通过分析智能车辆纵横向耦合系统的建模与控制方法,建立了智能车辆纵横向耦合运动学模型,该模型考虑纵向、横向以及横摆运动状态,可以更为精确地对智能车辆换道轨迹与换道完成后的车道保持进行控制。分析了车辆行驶过程中的纵横向耦合影响,提出了智能车辆纵横向耦合建模与控制问题,建立了智能车辆纵横耦合控制系统,包括纵向运动、横向运动以及横摆运动模型；采用指数型滑模变结构方法设计换道轨迹跟踪控制器,使智能车辆系统在车辆换道过程中满足期望的动静态性能指标；针对换道完成后的车道保持问题,建立智能车辆的横向偏差模型,采用Terminal滑模变结构方法设计车道保持控制器,将横向运动与横摆运动结合起来,使得横向偏差可以随着车道曲率变化而自动调节,同时,可以提高车道保持过程中纵横向运动的稳定性；采用MATLAB仿真工具对智能车辆换道轨迹跟踪控制以及换道完成后的车道保持控制进行仿真,验证了控制器设计的有效性和稳定性。针对某型越野车的结构特点,进行了智能车辆平台的机械改造,搭建了智能车辆的硬件和软件平台,并进行了实际高速公路的试验,验证了系统的可靠性和稳定性。根据智能车辆自主换道技术验证的需要,对原车进行了智能化改造,采用电动转向方案设计了车辆的转向机构,在原车制动系统中加装一套液压阀组改造了制动机构,通过并联安装一套电控拉线盘实现油门机构的智能控制；搭建了智能车辆的硬件平台,优化仪器设备；采用多线程技术进行软件设计,将系统分为主线程、控制线程、道路信息采集线程和串口通讯线程四个部分:进行了智能车辆高速公路自主驾驶试验,完成我国首次在权威机构、测试机构和新闻媒体三方共同监督下的智能车辆高速公路自主驾驶试验,测试公里数达到1500公里以上,共计完成自主换道95次,试验数据证明了智能车辆在高速公路环境下的自主换道具有较好的稳定性和可靠性。
[Abstract]:Intelligent vehicle has changed the traditional vehicle driving way fundamentally, the closed-loop system driver from "the driver of a vehicle - road of liberation, the use of advanced electronic and information technology to control the vehicle, that conventional driving activities, durable and easy fatigue operation automatically, can greatly improve the safety efficiency and personnel transportation system. The Research of intelligent vehicle autonomous switching technology, autonomous interaction can eventually realize multi vehicle cooperation, improve personnel, equipment efficiency and battlefield environment; at the same time, through the accurate perception of environmental information, and scientific and reasonable control algorithm of decision analysis and stability reliable, safety of the vehicle lane change ratio of uncertainty the lane change is more superiority and effective control of traffic accidents caused by human factors.
Through the process analysis of driver's driving behavior of driver, change and stage intention, analyzes the influencing factors of the lane change, and then in-depth study of the decision-making mechanism for the driver, and according to the structure characteristics of intelligent vehicle, decision process simulation of the lane change, analyzes the decision-making mechanism for autonomous intelligent vehicle Tao. Through the analysis of human organs in the process of driving the vehicle in the driver, the establishment of a system model of driver vehicle road, the driver was studied from the perspective of control theory, the function and role of vehicle and road in the system, and puts forward the evaluation index system control; through the experimental data. The main cause of lane change is a slow lane in front of the driver, and the pursuit of space and time are the main factors influencing the vehicle lane change; the car A lane change process is divided into the lane change intention, change the timing of the decision, changing path planning and trajectory tracking control in four stages; establish "highway typical scene driver - vehicle - road system, the driver's lane changing process of information perception to the driver behavior oriented, behavior decision and control process of the three modules and control decision interaction; analysis of the parameters characterizing the lane change decision stage, select the lane information, lane edge information, vehicle information such as the characteristic parameters of interactive intelligent vehicle autonomous lane change research; according to the structural characteristics of intelligent vehicle the lane change decision making process, simulation, system model is established for the" machine - vehicle - Road ", analyzes the decision-making mechanism of intelligent vehicle autonomous lane change.
Information fusion is carried out through the establishment of road radar map, lane change intention generation road in the process of change the timing of decisions, for trajectory planning of three stages analysis, set up the decision-making model of intelligent vehicle autonomous lane change, sudden abnormal situation and the change process of a static, obstacle avoidance the model of dynamic obstacles vehicle. Attention allocation mechanism simulation of human cognitive behavior, according to the concept of right of way, the establishment of the right size radar map, storage space and computing resources using less complete simulation of human cognitive behavior and counted by row radar map through information fusion, simulation analysis and path planning and other functions; the definition of the minimum safety distance of intelligent vehicle, based on the intelligent vehicle change intention. The value of expected tract sample of 2 skilled driver, select the typical highway road Line, make the test control conditions of design, characteristic parameters of impact of the acquisition of the lane change of the 311 groups, among them, lane keeping data of 97 groups, 214 groups of lane changing data; using v- training support vector machine, select (Delta, V) = (0.12,0.03) v- as the model parameters of support vector machine. The accuracy rate of lane changing is 91.05%; from the time of lane change, change lateral acceleration, curvature mutation analysis and comparison of the changing path planning method of intelligent vehicle used, according to the research object in this paper, the method of determining the trapezoidal acceleration changing path; the sudden and unexpected event change process, set up the static and dynamic model of vehicle obstacle avoidance; in order to meet the real-time requirements, design a variable scale grid in environment modeling, when the vehicle is traveling at a high speed variable scale grid than the traditional grid CPU occupancy is reduced about 34%; variable With the right scale grid radar map are fused by static selection and dynamic obstacle avoidance module obstacle avoidance strategy, to escape the vehicle sudden unexpected obstacles.
Through the analysis of intelligent vehicle longitudinal modeling and control method of coupling system, established intelligent vehicle longitudinal kinematic model, the model considering the longitudinal, lateral and yaw motion, can be more accurately on the changing path of intelligent vehicle and the lane change after the completion of the lane keeping control. Analysis of the vehicle in the vertical and horizontal coupling influence, proposed intelligent vehicle longitudinal modeling and control problems of coupling, established and coupling control system of intelligent vehicle, including longitudinal motion, lateral motion and yaw motion model; the exponential sliding mode variable structure design method for trajectory tracking controller, the intelligent vehicle system satisfies the desired dynamic and static performance the index in the vehicle lane change process; to change after the completion of the lane keeping, lateral deviation model of intelligent vehicle, using Terminal sliding mode variable structure The design method of lane keeping controller, the lateral motion and yaw motion combined with lateral deviation can make lane curvature change and automatic adjustment, at the same time, can improve the lane keeping process vertical movement stability; the intelligent vehicle lane changing trajectory tracking control and lane change after the completion of the lane keeping control using MATLAB simulation the simulation tool to verify the effectiveness of the controller design and stability.
According to the structure characteristics of a certain type of off-road vehicle, the mechanical transformation of intelligent vehicle platform, build intelligent vehicle platform of hardware and software, and test the actual highway is carried out and verified the reliability and stability of the system. According to the requirements of intelligent vehicle autonomous lane changing technology verification, the intelligent transformation of the original car the design of electric power steering, the steering mechanism of the vehicle, a hydraulic valve group reformed braking mechanism installed in the original brake system, by installing a set of parallel implementation of intelligent electric cable disc throttle body control; build intelligent vehicle hardware platform, optimization of equipment; software design adopts the multi thread technology the system is divided into the main thread, the thread of control, the four part of the road information acquisition thread and the serial communication thread: the intelligent vehicle autonomous highway driving test, I finished For the first time in the country authority, intelligent vehicle highway testing institutions and the media of all the three parties under the supervision of the autonomous driving test, test the number of kilometers to 1500 kilometers above, completed a total of 95 independent change, test data to prove that the autonomy in the highway environment for intelligent vehicle has good stability and reliability.

【学位授予单位】：中国人民解放军军事医学科学院
【学位级别】：博士
【学位授予年份】：2014
【分类号】：U463.6;U495

【参考文献】