社会与工程心理学研究室知识库

自动驾驶汽车(autonomous vehicle)的发展得到了各个国家及各大企业的重视，在自动驾驶汽车越来越多的进入市场的过程中，了解大众对于它的接受度非常重要的。但是自动驾驶给社会带来舒适、高效等诸多便利的同时，重大安全隐患特别是接管中的人机交互问题可能成为交通事故新的来源。探究如何设计人机交互系统，有效引导驾驶员从无关任务中转移并迅速完成对环境的感知，帮助驾驶员合理做出决策并准确完成动作对道路安全来说格外重要。因此本研究主要关注问题，一方面是探索终端消费大众对于自动驾驶汽车的接受度，另一方面探究如何设置自动驾驶车辆的提示预留时间有助于驾驶员在紧急情况下的处理。

研究一的目的是编制和验证中文版的自动驾驶接受度量表。本实验通过网络招募被试，邀请他们填写一系列问卷，包括自动驾驶接受度量表、交通氛围量表和人口学信息题目，最终共收集到有效问卷1453份，其中非驾驶员679份，驾驶员774份。实验结果发现，新编的自动驾驶接受度量表具有良好的内部一致性信度和结构效度;该量表共包括四个维度，分别是“系统有用性”、“场合优势”、“系统担忧”和“场景担忧”，它适合用来测量大众对于自动驾驶汽车的接受度，其中“系统有用性”和“场合优势”得分越高、“系统担忧”和“场景担忧”得分越低则代表其对自动驾驶汽车接受度越高;另外，交通氛围感知和是否拥有驾驶经验都会影响自动驾驶汽车接受度。驾驶员对自动驾驶汽车有更积极的态度，相较于非驾驶员，驾驶员对自动驾驶汽车的担忧要更少，并认为它们更有用。研究二的目的是使用驾驶模拟器探究不同接管场景和次要任务下的不同预留时间对接管行为的影响。本研究选取了复杂程度、危险程度以及紧急程度不同的三种场景，分别是前方出现减速带、前方出现事故车辆、前方车辆紧急刹车且相邻车道拥堵无法通行，次要任务分成1-back, 2-back和没有次要任务三组，预留时间选择3s, 4s, 5s, 6s和7s这五个连续的时间点，招募有效驾驶员被试87人在实验室进行驾驶模拟器的自动驾驶接管实验，研究结果发现，在前方出现减速带这种没有很大安全风险的情况下，4s的预留时间就足够驾驶员合理的判断道路状况并完成安全的接管;在前方出现事故车辆和前方出现紧急刹车这种容易出现碰撞事故的情况下，需要较长的预留时间即6s或7s才较为安全。本研究成果能够为自动驾驶汽车的接管系统设计提供丰富的理论和实证依据，提高自动驾驶汽车的安全性，有利于自动驾驶的推广，加快国家道路系统的升级。

The development of autonomous vehicles has been valued by various countries and major enterprises. In the process of more and more self-driving cars entering the market, it is vital to understand end-users' acceptance of AVs. While automatic driving brings comfort, efficiency and many other benefits to the society, major safety hazards, especially the human-computer interaction problems in the process of take-over, has become a new source of serious accidents. Exploring how to design human-computer interaction systems is particularly important for road safety. A good system could effectively guide drivers to transfer their attention from irrelevant tasks to driving tasks including perceive the surrounding environment rapidly. Thus, it helps drivers make reasonable decisions and accurately take control of vehicles. The main concerns of present research are to explore the acceptance of AVs among Chinese end-users' and to explore the time budget design of the automated driving take-over interface.

The purpose of the first study was to develop and validate a questionnaire to assess public receptivity toward AVs. A total of 1453 participants (679 non-drivers and 774 drivers) voluntarily and validly completed a series of questionnaires. The questionnaires included the Autonomous Vehicle Acceptability Scale (AVAS), the Traffic Climate Scale (TCS), and sociodemographic variables. The satisfactory internal consistency reliability and construct validity revealed that the newly developed Chinese version of the AVAS is a suitable tool to measure public acceptance of AVs. Moreover, exploratory factor analysis and confirmatory factor analysis conformed to the four factors of AVAS, including benefits in usefulness (BILT), benefits in situations (BIS), concern scenarios (CS) and system concern (SC). Scores higher on benefits and lower on concerns represent more acceptance of AVs. In addition, we found that the public's perceived local traffic safety climate affected the attitude toward AVs. Furthermore, the differences between drivers and non-drivers revealed that drivers were concerned significantly less about AVs and regarded AVs as more useful than non-drivers did. The purpose of the second study was to explore the effects of different time budget on take-over performance in different take-over scenarios and non-drive related tasks (NDRT). A total of three scenarios were selected which were different in complexity, degrees of danger and urgency are explored. These are the scenes with the first scenario was a speed bump in front, the second scenario was two accident vehicles blocked in front and the third scenario was a vehicle suddenly brake in front with traffic jam in the adjacent lane. Two groups were engaged in 1-back and 2-back task while turn on the automated system respectively and one group didn't do any task. Five time budgets were 3s, 4s, Ss, 6s and 7s. We totally recruited 87 drivers for the experiment. According to the results, the 4s time budget is enough for the driver to reasonably judge the road condition and complete the safe take-over when facing the scenario without large security risk; however, when facing the scenario which are prone to collision accidents, it takes a longer time budget, such as 6s or 7s, to be safer. The results of the research would provide abundant theoretical and empirical basis for the design of automated driving take-over systems, improve the safety of automated driving vehicles, facilitate the promotion of automated driving, and accelerate the upgrading of the national road system.