基于可达集的自主车辆超车决策  

Autonomous Vehicle Overtaking Decision Based on Reachable Sets

作　　者：韩文奇 杨旭 校艳 党薪艳 钟帅 Han Wenqi;Yang Xu;Xiao Yan;Dang Xinyan;Zhong Shuai(Xi’an Traffic Engineering Institute,Xi’an,710300,China)

机构地区：[1]西安交通工程学院,西安710300

出　　处：《黑龙江科学》2023年第14期14-18,共5页Heilongjiang Science

基　　金：大学生创新创业训练计划项目(2022DC07)。

摘　　要：车辆在自主驾驶中会遇到随机障碍物,存在无法识别问题,且自主驾驶的安全性验证模型大多为单个目标,导致自主驾驶车辆系统的适应性能力不高。将哈密顿雅可比方程与可达集分析方法相结合,提出在自主驾驶超车决策模型中对即将进行超车的车辆同向道增加一辆目标车辆及反向车道增加一辆目标车辆的方法,以可达集为约束条件改进模型,得出自主驾驶车辆与目标车辆在任何环境下的行为。采用水平集等方法来描述自主驾驶车辆的超车过程,结合自主驾驶车辆在不同环境下可能会发生的不同运行状态,引入曲线道路下车辆行驶过程,得出车辆行驶的特性曲线函数及自主驾驶车辆在不同行驶状态的下可达集变化。结果表明,此方法在交通堵塞道路及超车环境下,能够评估自主驾驶车辆的安全性,为车辆的智能驾驶决策提供参考。Autonomous vehicle encounters a problem such as a random obstacle in autonomous driving.At the same time,the safety verification model of the vehicle in autonomous driving is mostly a simple case of a single target,and the adaptability of the self-driving vehicle system is not good.By combining Hamiltonian Jacobi equation with reachability set analysis method,a method of adding one target vehicle in the same lane and one target vehicle in the opposite lane to the autonomous driving overtaking decision model is proposed.The model is improved with reachability set as the constraint condition,and the behavior of the autonomous driving vehicle and the target vehicle in any environment is obtained.The overtaking process of autonomous vehicles is described by level set and other methods,and the driving process of autonomous vehicles under curvilinear road is introduced in combination with the different running states of autonomous vehicles under any environment.The characteristic curve function of vehicle driving and the change of accessibility set of autonomous vehicles under different driving states are obtained.The results show that the method can be used to evaluate the safety of autonomous driving vehicles in traffic jams and overtaking environment,and provide reference for intelligent driving decisions of vehicles.

关 键 词：安全性验证 智能驾驶 可达集分析 自主驾驶超车决策 

分 类 号：O225[理学—运筹学与控制论] U463.6[理学—数学]