【作者】 吴乙万；

【导师】 钟志华； 黄智；

【作者基本信息】 湖南大学 ， 车辆工程， 2013， 博士

【摘要】 随着社会的进步和经济的发展，汽车已成为人类生活不可缺少的一部分。汽车为人类文明进步和社会经济发展做出巨大贡献的同时，汽车事故也呈上升趋势。道路交通安全已成为全球性的社会问题，提高车辆安全性成为研究热点。论文以车道偏离辅助驾驶系统为研究对象，对其中部分关键技术进行了研究。本文研究了车道偏离辅助驾驶决策机制，从主客观方面对车道偏离预警人机接口进行评价，研究了车道偏离辅助控制方法，建立了车道偏离预警与辅助控制的快速原型试验台，通过快速原型对比试验及驾驶员在环台架试验对本文所提出的方法进行验证。提出了基于动态TLC阈值的车道偏离联合决策机制。针对传统TLC决策模型因固定的触发阈值可能导致误警/误干预或漏警/漏干预的问题，综合考虑车路航向角偏差、路面附着系数、车速及执行机构响应时间等对车辆姿态校正过程的影响，建立了动态TLC阈值的计算模型。决策机制自动调节预警和辅助控制的触发时机，避免过早介入对驾驶员产生不必要的干扰，同时避免过晚介入导致偏离车道。从感知—决策—执行的层次角度出发，建立了人机协同的车道偏离辅助系统的控制体系结构。研究驾驶员优先决策下的预警—辅助控制的人机协同决策模型及其实现方法。在Carsim/simulink联合仿真平台上验证了基于动态TLC阈值的车道偏离辅助驾驶决策方法有效性和正确性。设计了车道偏离预警人机接口的主客观评价指标和评价方法，基于自主研制的驾驶模拟器进行了车道偏离预警对比试验，获得了车道偏离预警的较优人机接口方式。针对现有的研究专注于各种预警人机接口的预警效果忽略驾驶员主观感受的不足，进行车道偏离预警台架试验，根据提出的主客观指标评价系统，采用统计方法系统分析了驾驶员对听觉、视觉和触觉等预警人机接口方式的敏感程度及可接受度。获得较优的车道偏离预警人机接口方式。通过台架试验进一步分析触觉预警的波形、幅值和频率等参数对预警效果及可接受度的影响，给出了改善触觉预警效果及可接受度的有效途径。根据分层控制思想，研究了基于差动制动的车道偏离辅助控制方法。在辅助驾驶决策的基础上，基于预瞄点处的车-路偏差、车辆状态和道路附着限制计算期望的横摆响应，设计滑模控制器控制辅助横摆力矩，通过差动制动产生横摆力矩，使车辆横摆响应跟踪目标值，达到车道偏离辅助控制的目的。提出了轮缸压力闭环控制和滑移率均衡方法实现横摆力矩伺服控制。在轮缸压力闭环控制方法中，分别采用制动力定比例分配方法与考虑轮胎工作状态的制动力最优控制分配。在滑移率均衡控制方法中，提出等效前轮滑移率的概念，根据期望横摆力矩及横摆角速度偏差选取制动轮，根据等效前轮滑移率和后轮滑移率提出了滑移率均衡策略，进行制动力次优分配。采用模糊控制器设计底层伺服，产生辅助横摆力矩，达到使车辆横摆响应跟踪期望值的目的。在Carsim/simulink联合仿真平台上对基于差动制动的车道偏离辅助控制方法进行仿真试验，研究结果表明所提出方法将车辆限制在车道范围内，有效避免车道偏离事故发生。建立了由CarsimRT/LabviewRT实时平台及转向、制动执行机构硬件组成的车道偏离辅助驾驶系统快速原型试验台架，对基于差动制动的辅助控制方法进行台架试验，其结果与Carsim/simulink联合仿真结果基本一致。各种工况的台架试验表明，采用动态TLC阈值能适时触发预警与辅助控制，基于控制分配算法的制动力分配方法具有更好的鲁棒性。滑移率均衡方法较压力闭环方法控制粗糙，但对控制对象模型参数适应性好，无需额外传感器且简单可靠。本文的创新点如下：(1)提出基于动态TLC阈值的车道偏离联合决策机制，该决策机制能根据车路航向角偏差、路面附着及执行机构响应等进行动态决策，适时触发预警与辅助控制。(2)基于台架试验，研究得出满足车道偏离预警效果且易被驾驶员接受的较优预警人机接口方式及改善途径。(3)设计了基于传统ABS/ESP系统实施差动制动的两种伺服控制方法，实现车道偏离时的车辆纵横向控制。具有对驾驶员的干扰小、人机协同策略简单、可靠性高、工程实用和低成本的特点。更多还原

【Abstract】 With the development of society and economy, vehicles have become an indispensable part of human life. Although vehicles make a significant contribution to the progress of human civilization and socio-economic, vehicle accidents are on the rise. Therefore, road traffic safety has become a global social problem. How to improve vehicle safety is currently a research hotspot. This paper focused on the lane departure assistance system(LDAS), and studied the key technologies of LDAS.A new strategy for the decision of lane departure avoidance was proposed. The performance of lane departure warning techniques was evaluated with the consideration of all kinds of objective and subjective factors. A novel approach for lane departure avoidance was presented, which applies additional yaw moment to the vehicle to correct lane departure of the vehicle by means of differential braking. A rapid prototyping test platform with the function of lane departure warning and assistance control was established. The proposed methods were validated on the platform and driver-in-loop test.The main contents of this paper are as follows:(1) A new strategy for the joint-decision of lane departure avoidance, which is based on the dynamic threshold of Time to Lane Crossing (TLC) was presented. Fixed threshold may lead to false alarm or missed alarm, even false intervention or missed intervention. A model of the dynamic threshold of TLC was proposed, which is calculated by vehicle-lane heading, vehicle speed, road adhesion and reaction time of the implementing agencies and driver in real-time. Too early intervention may produce unnecessary interference to the driver, while too late intervention may lead to lane departure. The proposed decision strategy can adjust the threshold for warning and control. A human-machine cooperation control architecture was proposed with the consideration of perception, decision-making and execution. The proposed strategy was validated in a Carsim/Simulink co-simulation platform.(2) Objective and subjective assessment criteria for lane departure warning were designed. A series of lane departure warning experiments were conducted based on a more realistic fixed-based driving simulator. In previous studies, researchers placed more emphases on the efficiency of warning system instead of driver’s subjective feeling. In this study, subjective and objective evaluation indexes were proposed. Effectiveness and drivers’ acceptance of seven types of warning techniques were further studied through a series of in-door experiments. Statistical methods were used to systematically analyze drivers’ sensitive to the auditory, visual and haptic warning technologies. A solution of warning technique to improve the effectiveness and acceptance was brought forward. The effects of pattern, amplitude and frequency of haptic warning on the effectiveness and drivers’ acceptance were studied through another series of in-door experiments. A method to improve the performance and drivers’ acceptance of haptic warning was put forward.(3) A novel approach for lane departure avoidance was proposed, which applies additional yaw moment to the vehicle to correct lane departure of the vehicle by means of differential braking. The desired yaw rate is determined by considering vehicle-lane deviation, vehicle dynamic and limitation of road adhesion. A sliding mode controller (SMC) was designed to control the additional yaw moment. The yaw moment is produced by the distribution of brake forces between the same side brakes, and lane departure avoidance is carried out by tracking the desired yaw response. In order to produce the additional yaw moment, a brake pressure close-loop control and a slip ratio balance servo control methods were presented. In terms of pressure close-loop control, fixed ratio and control allocation were used as the assignment methods for braking force distribution respectively. In terms of slip ratio balance control, a concept of equivalent slip ratio was put forward. The selection of the brake wheels is according to the additional yaw moment and yaw rate deviation. A new strategy for the balance of the equivalent slip ratio and rear wheel slip ratio was presented. The brake force is distributed in suboptimal. Two fuzzy controllers were designed to produce the additional yaw moment, and lane departure avoidance was carried out by tracking the desired yaw response. The proposed Lane departure avoidance method was evaluated in Carsim/simulink. Simulation results show that the proposed methods can restrict vehicle into the lane range and avoid lane departure accident.(4) A rapid prototyping test platform for lane departure driving assistance system was established, which is composed of CarsimRT/LabviewRT real-time platform, real steering and braking system. The methods, which were proposed in this study, are validated on the platform. The test results are conisistent with the simulation results. Several conclusions are obtained from various driving conditions. Dynamic TLC threshold can timely trigger warning and control. Brake force distribution method based on control allocation has better robustness. The slip ratio balance method has a good adaptability to the model parameters, although the control results are rougher than the braking pressure close-loop control method. The slip ratio balance method doesn’t need any additional cost, and is simple and reliable.The innovations of this paper are as follows:(1) A model of the dynamic threshold of TLC was presented, which is calculated by vehicle-lane heading, vehicle speed, road adhesion and reaction time of the implementing agencies and driver in real-time. The dynamic TLC threshold can trigger warning and control more timely.(2) Based on the in-door experiments, a solution of warning technique to improve the effectiveness and acceptance was brought forward.(3) A novel approach for lane departure avoidance was studied and designed, which applies additional yaw moment to the vehicle to correct lane departure of the vehicle by means of differential braking. The proposed control method has the following characteristics: less interference to driver, simple strategy of human-machine cooperate control, high reliability, practical and low cost.更多还原