【作者】 李冬；

【导师】 陈静；

【作者基本信息】 吉林大学 ， 车辆工程（专业学位）， 2020， 硕士

【摘要】 随着人们对道路交通安全重视程度的增加,汽车系统也开始集成越来越多的主动安全技术。目前对于纵向防碰撞系统的研究大多是采用预警和制动操作来规避碰撞风险,但是在一些速度较快的紧急工况中,如前方障碍车突然停止、左右侧的视野盲区经过行人等,这类系统由于制动距离较长,作用效果十分有限,针对纵向防碰撞系统局限性,本文提出了一种快速行驶下通过转向介入,改变汽车的横向位置以达到避障效果的紧急转向算法。首先,本文基于国内外主动避障技术的研究现状,设计了紧急换道转向避障系统的总体架构,明确了各个功能模块之间的信息交互。按照紧急程度将主动避障行为分为碰撞报警,制动避障,和紧急转向避障三种形式,针对各个形式的特点,本文基于碰撞时间的倒数确定主车的安全状态,若存在碰撞风险,则进一步根据安全距离决策危险等级,按照危险等级系数采取不同的操作,同时,考虑到相关法律和驾乘人员安全性,本文设计了驾驶员接管策略以分配车辆的控制权。其次,本文分析了紧急换道转向避障应用场景的特点对换道轨迹提出了五点要求,由于传统的几种紧急换道轨迹规划方法并未考虑动态障碍物以及周围其他交通车等影响,所以本文选择了基于数值优化的规划方法,在紧急换道避障系统执行的有限时域内,对障碍车辆以及换道侧车辆未来的运动状态进行数学建模,根据各个车辆的空间位置决策纵向的可行驶区域及行驶轨迹,若存在可行的纵向轨迹,再根据障碍车横向位置、车道线边界以及换道宽度等决策横向可行驶区域及行驶轨迹,最终得到一系列离散时域内的期望运动状态集合,由于纵向和横向的最终轨迹是通过目标函数最优求解得到的,所以在避障的基础上尽可能的保证了驾乘人员的舒适性,同时,为了更好地评估安全状态,本文还提出了一种碰撞检测方法,检测规划轨迹发生碰撞的可能性。然后,针对规划轨迹的特性,本文将轨迹跟踪算法解耦为速度跟随控制和横向位移跟踪控制,针对速度跟随控制,本文设计了模糊自整定的比例-积分控制器,针对横向位移跟踪控制,本文设计了基于前馈-反馈、基于模型预测两种控制器并且通过不同速度下的双移线轨迹测试这两种控制器的效果。最后,本文基于CarSim/Simulink联合仿真测试了不同工况下紧急换道转向避障规划算法、速度跟随算法以及基于模型预测的横向位移跟踪算法耦合的效果,为了验证实车环境下算法的功能性,本文对某国产SUV进行了升级改装,建立了实车测试平台,同时考虑紧急换道实车测试的危险性,本文对实际的换道环境进行了简化,测试了在高速公路场景下,规划算法和基于前馈-反馈的横向位移跟踪控制器在一般换道行为中的适应性。更多还原

【Abstract】 As people attach more and more importance to road traffic safety,more and more active safety technologies are integrated into automotive systems.At present,most of the obstacle avoidance systems are using warning and braking operations.However,in some emergency situations,such as the sudden stop of the front vehicle and pedestrians out of blind spot area,etc,due to the long braking distance,the effect is very limited.Aiming at the limitations,this paper proposes an emergency steering algorithm that changes the vehicle’s lateral position to avoid obstacles during fast driving.Firstly,based on the current status of obstacle avoidance technology all around the world,this paper designs the overall architecture of the emergency lane change steering obstacle avoidance system,and the information interaction between the various functional modules is clarified.According to urgency,obstacle avoidance behavior is divided into three forms: collision warning,emergency braking,and emergency steering obstacle avoidance.the safety status of the ego vehicle is based on the reciprocal of the collision time in this paper.If there is a risk,the danger level is further determined based on the safety distance,According to the danger level,different operations are taken.At the same time,considering the relevant laws and the safety of drivers and passengers,this paper designs a driver takeover strategy to allocate control of the vehicle.Secondly,this paper analyzes the characteristics of the application scenario of emergency lane change,and propose five requirements for the lane change trajectory.Because the traditional several emergency lane change trajectory planning methods do not consider the impact of dynamic obstacles and other surrounding vehicles,this paper chooses a planning method based on numerical optimization.In the limited time domain,the emergency lane change obstacle avoidance system mathematically model the future state of the obstacle vehicle and others,then determine the longitudinal drivable area and driving trajectory.feasible longitudinal trajectory exists,the laterally drivable area and driving trajectory are determined according to the obstacle vehicle,the lane boundary,and the lane change width,etc.,and finally a series of expected motion state sets in the discrete time domain are obtained.Since the final trajectories are obtained by the optimal solution of the objective function,the comfort of the driver is guaranteed on the basis of obstacle avoidance.At the same time,in order to evaluate the safety status,A collision detection algorithm is proposed to detect the possibility of collision of planned trajectories.Then,according to the characteristics of the planned trajectory,this paper decouples the trajectory tracking algorithm into speed following control and lateral displacement tracking control.For speed following control,a fuzzy self-tuning proportional-integral controller is designed.For lateral displacement tracking control,this paper designs two controllers based on feedforward-feedback or model prediction,and double lane trajectory at different speeds tests the effect of these two controllers.Finally,based on the CarSim/Simulink co-simulation,this paper tested the coupling effects of the trajectory planning,speed following,and lateral displacement tracking algorithm based on model prediction.In order to verify the functionality of the algorithm in a real environment,this paper has upgraded a SUV and established a real vehicle test platform.At the same time,considering the danger of the real vehicle test,this paper simplified the actual lane changing environment,and tested the general lane changing behavior of the planning algorithm and the lateral displacement tracking controller based on feedforward-feedback in the highway scene.更多还原