摘要
针对自动驾驶汽车在标准道路下换道避撞的场景,设计基于5次多项式的多段式避撞路径,根据换道初始与结束时的边界条件确定5次多项式的各项系数,以侧向的安全约束确定纵向的换道安全距离,并考虑换道路径的最大侧向加速度,保证了自动驾驶避撞过程中的舒适性。设计模型预测控制器,采用线性时变模型,将车辆动力学模型线性化后离散化,设计的目标函数中包括跟踪误差和控制增量,保证自动驾驶汽车可以在完成避撞路径跟踪的同时,消耗最少的能量。添加了控制量约束、控制增量约束和输出约束,使得所设计的路径跟踪控制器满足自动驾驶汽车执行机构物理约束的要求。最后通过建立Matlab和CarSim联合仿真平台,验证了不同工况下所设计的避撞路径规划与跟踪控制系统的有效性。
Aimed at the scene of autonomous vehicle changing lanes to avoid collision on standard roads,we plan the collision avoidance path of autonomous vehicle through quintic polynomial and determine the coefficients of quintic polynomial according to the boundary conditions at the beginning and end of lane changing.We also determine the longitudinal lane changing safety distance with lateral safety constraints and consider the maximum lateral acceleration of lane changing path,ensuring the comfort in the process of automatic driving collision avoidance.Then we design the model predictive controller.We use the linear time-varying model to linearize the vehicle dynamics model and then discretize it.The designed objective function includes tracking error and control increment to ensure that the autonomous vehicle can complete the collision avoidance path tracking while consuming the least energy.We add the control quantity constraint,control incremental constraints and output constraints,enabling the designed path tracking controller to meet the physical constraints of the autonomous vehicle actuator.Finally,by establishing the joint simulation platform of Matlab and CarSim,we verify the effectiveness of the collision avoidance path planning and tracking control system designed under different working conditions.
作者
杨航
张成涛
覃立仁
赵浙栋
YANG Hang;ZHANG Chengtao;QIN Liren;ZHAO Zhedong(School of Mechanical and Automotive Engineering,Guangxi University of Science and Technology,Liuzhou 545616)
出处
《广西科技大学学报》
CAS
2023年第3期7-13,28,共8页
Journal of Guangxi University of Science and Technology
基金
国家自然科学基金项目(52202491)
广西创新驱动发展专项资金项目(桂科AA19182006-2)
广西科技大学博士挂职驻柳企业科研项目(BSGZ2109)资助。
关键词
自动驾驶
路径规划
5次多项式
模型预测控制
automatic driving
path planning
quintic polynomial
model predictive contro
