无人驾驶汽车决策系统的规则正确性验证

发布时间：2018-06-14 13:57

本文选题：无人驾驶车辆 + 决策系统　；参考：《国防科学技术大学》2015年硕士论文

【摘要】：随着信息技术的发展,无人驾驶车辆技术日渐成为科学研究的重点领域之一,各国都在其中投入了大量的人力物力。无人驾驶车辆技术能够将人类驾驶员从枯燥危险的驾驶工作中解放出来,减少了车辆行为的随机性,使车辆的行为变得可预测,可以极大地提高车辆行驶的稳定性。另外,无人驾驶车辆技术提高了车辆对环境的反应速度,增强了车辆的安全性,也可以大大缩短行车间距,从而增加公路的运输能力。同时,无人驾驶车辆技术改善了车辆对环境的感知精度,消除了因驾驶员个人问题而造成的交通事故。无人驾驶车辆技术的核心在于其决策系统的开发,当前无人车决策系统开发过程中面临着诸多难题,主要包括以下几点:由人工编写的代码成本高且维护困难;决策规则与系统软件没有实现分离;设计过程中可能存在潜在的缺陷和错误;软件开发与安全检查不同步。针对以上问题,课题组提出了验证驱动的基于代码自动生成的无人车决策系统开发框架。课题组早期的工作包括设计了描述无人车决策规则的中间语言,并设计实现了无人车决策系统辅助开发工具,解决了开发过程中的前两个难题。为了解决无人车决策系统中的安全性问题,本文在代码自动生成技术的基础上引入了模型检验技术,自动生成无人车的模型代码,并对无人车决策系统进行环境建模,通过形式化验证可以发现无人车决策系统设计过程中不易察觉的缺陷和错误,解决其安全性不足的问题,同时能够将安全检查与软件开发过程同步,降低其维护成本。基于该框架,本文将模型检验模块集成到已有的无人车决策系统辅助开发工具UNMANNED_RULE_EDIT(URE)中,为无人车决策系统的开发工作提供帮助,为其安全性提供保障。
[Abstract]:With the development of information technology, driverless vehicle technology has increasingly become one of the key areas of scientific research, in which countries have invested a lot of manpower and material resources. Driverless vehicle technology can liberate human drivers from the boring and dangerous driving work, reduce the randomness of vehicle behavior, make vehicle behavior predictable, and greatly improve the stability of vehicle driving. In addition, the technology of driverless vehicles can improve the response speed of vehicles to the environment, enhance the safety of vehicles, and greatly shorten the distance between vehicles, thus increasing the transportation capacity of highways. At the same time, the driverless vehicle technology improves the environmental perception accuracy of the vehicle and eliminates the traffic accidents caused by the driver's personal problems. The core of driverless vehicle technology is the development of its decision-making system. At present, there are many difficulties in the development of unmanned vehicle decision system, including the following: the cost of manual code is high and the maintenance is difficult; Decision rules are not separated from system software; there may be potential defects and errors in the design process; and software development and security checking are out of step. In view of the above problems, the research group proposed a verification driven development framework for unmanned vehicle decision system based on automatic code generation. The early work of the research group includes the design of an intermediate language to describe the decision-making rules of the unmanned vehicle, and the design and implementation of an assistant development tool for the decision-making system of the unmanned vehicle, which solves the first two difficult problems in the development process. In order to solve the security problem in the unmanned vehicle decision system, this paper introduces the model checking technology on the basis of the code automatic generation technology, automatically generates the model code of the unmanned vehicle, and carries on the environment modeling to the unmanned vehicle decision system. Through formal verification, we can find the imperceptible defects and errors in the design process of the unmanned vehicle decision system, solve the problem of insufficient security, synchronize the security inspection with the software development process, and reduce the maintenance cost. Based on this framework, the model checking module is integrated into Unmannet\
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U463.6;TP311.52

【参考文献】