Effect of instructing system limitations on the intervening behavior of drivers in partial driving automation
Effect of instructing system limitations on the intervening behavior of drivers in partial driving automation
复制标题
部分驾驶自动化中指令系统限制对驾驶员干预行为的影响
DOI：
10.1007/s10111-019-00568-1
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
S. Kitazaki
中科院分区：
文献类型：
--
作者：
Huiping Zhou;M. Itoh;S. Kitazaki
正确理解自动化的局限性对于驾驶员处理意外的紧急情况至关重要。本研究的重点是对不具备使用驾驶自动化系统知识和经验的新手驾驶员的限制进行基于解释的知识。讨论这些知识时考虑了（1）当驾驶自动化无法处理这种情况时，自动化无法发出警报的可能性；（2）从功能或景观的角度描述限制的方式。在驾驶模拟器中进行了 2 × 2 个解释性知识条件下的实验（每个条件 n = 24 名参与者，参与者的平均年龄 = 55.9 ± 16.2 岁）。收集有关从自动控制到手动控制的过渡时间的数据。结果表明，与功能描述（88%，2.3 秒）相比，如果从场景角度描述知识（平均安全干预率 = 95%，平均反应时间 = 2.0 秒），驾驶员可以更安全地进行干预。人们发现，在涉及系统限制的情况下，显式/场景知识对于响应警报以及处理关键系统故障更加有益。对扫视行为和访谈的进一步调查表明，具有显性/功能性知识的新手驾驶员容易过度依赖自动化的功能。因此，本研究阐明，在给予驾驶员明确的系统限制/故障知识的同时，对于感知和响应系统限制以及系统故障导致的意外危险具有指导意义。
Proper understanding of automation limitations is vital in order for drivers to deal with unexpected critical situations. The present study focuses on explanation-based knowledge of the limitations given to novice drivers who have no knowledge and experience of using driving automation systems. The knowledge is discussed considering (1) the possibility of the automation failing to issue an alert when the driving automation cannot handle the situation and (2) the manner of describing the limitations from either a functional or scenic point of view. An experiment conducted under 2 × 2 conditions of explanation-based knowledge ( n = 24 participants per condition, average age of the participant = 55.9 ± 16.2 years) is implemented in a driving simulator. Data on transition time from automated control to manual control are collected. The results reveal that drivers could intervene more safely if the knowledge is described from a scenic point of view (average ratio of safe intervention = 95%, average reaction time = 2.0 s), as compared to a functional description (88%, 2.3 s). Explicit/scenic knowledge was found to be more beneficial in responding to alerts in such situations involving system limitations as well as in dealing with critical system failures. Further investigation of glance behavior and interviews revealed that novice drivers with explicit/functional knowledge are prone to be over-reliant on the automation’s capability. Therefore, the present study clarified that providing a driver with knowledge about system limitations/failures explicitly while giving is instructive for perceiving and responding to system limitations as well as unexpected hazards due to system failures.