摘要
由于自主车辆具有离散控制和连续行为混成以及不可避免的系统偏差的特点,使传统验证方法无法遍历所有不确定的系统输入和运行状态,因而自主车辆安全决策的置信水平较低。为此,提出以车道边界作为约束条件的多目标约束的可达集建模方法,该方法利用哈密顿雅可比方程控制可达集变化,利用水平集完成可达集的可视化任务,确定具有不确定系统输入和运行状态的自主车辆安全区域。结果表明,与传统仿真方法相比,基于可达集的自主车辆多目标安全决策方法具有更高的适应性,并且可以为智能驾驶车辆在较为复杂环境下的安全性提供理论保证。
Due to the autonomous vehicles have the characteristics of discrete control and continuous behavior blending and inevitable system deviation,the traditional verification method can not traverse all uncertain system input and operation State,so the confidence level of autonomous vehicle safety decision is low. For this reason,a multi-objective constraint model method based on lane boundary is proposed,which uses Hamiltonian Jacobian equation to control the adjustable set change,uses the horizontal set to complete the visualization task of the Reach set,and determines the safe area of autonomous vehicles with uncertain system input and operation state. The results show that compared with the traditional simulation method,the multi-objective safety decision method of autonomous vehicle based on Danone has higher adaptability,and can provide theoretical guarantee for the safety of intelligent driving vehicle in more complex environment.
引文
1刘秀芳,曹凯,杨旭,等.基于可达集的自主车辆安全性验证方法[J].广西大学学报,2018,43(2):488-497Liu Xiufang,Cao Kai,Yang Xu,et al. Safety verification method of autonomous vehicle based on reachable set[J]. Journal of Guangxi University,2018,43(2):488-497
2 田晶晶,李世武,孙文财,等.基于车路协同的动态车辆安全状态评价模型[J].哈尔滨工业大学学报,2012,44(4):132—135Tian Jingjing,Li Shiwu,Sun Wencai,et al. Dynamic vehicle safety evaluation model based on vehicle-road coordination[J]. Journal of Harbin University of Technology,2012,44(4):132-135
3 周华.基于车-路条件的车辆安全行驶速度仿真研究[D].重庆:重庆交通大学,2012Zhou Hua. Simulation study on vehicle safe driving speed based on vehicle-road condition[D]. Chongqing:Chongqing Jiaotong University,2012
4 Albert W,Jonathan P H. Guaranteed infinite horizon avoidance of unpredictable dynamically constrained obstacles[J]. Autonomous Robots,2012,33:227-242
5 Jaeyong Park B S. Safe controller design for intelligent transportation system applications using reachability analysis[D]. Columbus:Ohio State University,2013
6 曹凯,刘秀芳,杨旭,等.自主车辆动态冲突控制可达集建模方法[J].系统仿真学报,2017,29(8):1658-1666Cao Kai,Liu Xiufang,Yang Xu,et al. Reachability set modeling method for autonomous vehicle dynamic conflict control[J]. Journal of System Simulation,2017,29(8):1658-1666
7 曹凯,黄肖肖,于云,等.自主车辆避障安全路径的可达集建模[J].系统仿真学报,2016,28(3):526-533Cao Kai,Huang Xiaoxiao,Yu Yun,et al. Reachability set modeling of obstacle avoidance safety path for autonomous vehicles[J]. Journal of System Simulation,2016,28(3):526-533
8 刘秉政,曹凯.自主车辆行为决策的安全验证方法[J].山东理工大学学报(自然科学版),2011,25(6):6-12Liu Bingzheng,Cao Kai. A safety verification method for autonomous vehicle behavior decision making[J]. Journal of Shandong University of Technology(Natural Science),2011,25(6):6-12
9 Mitshell I M. A toolbox of level set methods[D]. Vancouver:University of British Columbia,2004
10 方敏,张雅顺,李辉.混合系统的形式验证方法[J].系统仿真学报,2006,18(10):2921-2924Fang Min,Zhang Yashun,Li Hui. Formal verification method for hybrid systems[J]. Journal of System Simulation,2006,18(10):2921-2924
11 甘庭,夏壁灿.运用栅栏函数验证连续系统的有界时间安全性[J].软件学报,2016,27(3):645-654Gan Ting,Xia Bican. The bounded time security of continuous system is verified by using fence function[J]. Journal of Software,2016,27(3):645-654
12 Nazin A,Nazin S,Polyak B. On convergence of external ellipsoidal approximations of the reachability domains of discrete dynamic linear systems[J]. Automation and Remote Control,2004,65(8):1210-1230