【摘要】 自动驾驶（AD）技术是一种能够增强或替代人类驾驶员的技术，但同时也给汽车测试和评估（T&E）技术带来了极大的挑战。由于AD技术替代了驾驶员的感知、决策和控制，因此自动驾驶汽车（AVs）的T&E程序必须从独立的人-车二元T&E转变为人-车系统的集成T&E。在复杂的天气、交通和电磁环境下进行公路测试，使得AVs的定量T&E非常困难。为了解决这个问题，本文提出了一种多维度评价准则方案，其中包括运行范围、性能表现和交互与风险保障评价准则。通过形式化方法，即线性时间逻辑（LTL），实现了基于场景的评估方法。在中国GB的典型情景中，本文提出的评价准则方案得到了验证，并且通过基于SCANeR studio软件的模拟实现。更多还原

【Abstract】 Autonomous driving(AD) is an enhancement or replacement for human drivers, while AD has brought a huge challenge to automotive testing and evaluation(T&E) technologies. As AD replaces drivers’ perception, decision-making and control, the T&E procedure for autonomous vehicles(AVs) changes from independent T&E of human-vehicle binary to integrated T&E of human-vehicle systems. Complex weather, traffic and electromagnet environments under open-road testing make a quantitative T&E of AVs very difficult.To solve this problem, in this paper, we propose a multi-dimensional evaluation criterion scheme for AVs, including evaluation criteria for operational scope, behavioral performance, interaction and risk protection. Then, a scenario-based evaluation approach is realized by the formal approach, i.e., linear temporal logic(LTL). The proposed evaluation criterion scheme is verified by a typical scenario in China GB and implemented by simulations based on the SCANeR studio software.更多还原