Faculty of Engineering, Information and Systems, University of Tsukuba, Tsukuba, Japan; Japan Automobile Research Institute, Autonomous Driving Research Division, Tsukuba, Japan.
Graduate School of Systems and Information Engineering, Department of Risk Engineering, University of Tsukuba, Tsukuba, Japan.
Accid Anal Prev. 2022 Sep;174:106719. doi: 10.1016/j.aap.2022.106719. Epub 2022 Jun 2.
This study analyzes the impact of cooperative human-machine interface designs on drivers' trust in and interaction with automated driving systems (ADSs) during lane changes on highways. While drivers' inappropriate trust in ADSs can affect their behavior toward the system, capability to detect inadequate system performance, and perception of surrounding traffic disturbances, their engagement in the automated process can improve their comprehension of the system and traffic conditions, which is necessary for the safe practice of automated driving. Forty drivers practiced conditional driving automation in a driving simulator and encountered traffic congestion on the main lane of a two-lane highway. Four ADS designs were proposed to bypass the congestion. ADS-1 detects the congestion and synchronizes the speed accordingly. ADS-2 requests the driver to resume manual control and overtake the congestion. ADS-3 requests the driver to push a button to let the system overtakes the congestion automatically. ADS-4 overtakes the congestion automatically after informing the driver, while the driver can cancel it by pushing a button within 6 s. In all these conditions, driver intervention was optional. Although the drivers preferred and trusted ADS-1 and ADS-2 more than ADS-3 and ADS-4, the results indicate significant improvements in the driving performance and system usage under ADS-3 and ADS-4. Driving with ADS-3 improved drivers' engagement and reduced the requirement for control transfer compared with other systems. However, the time headway between the subject and adjacent vehicles indicated that lane changes were more critical under ADS-3 and ADS-4 than ADS-1 and ADS-2. Such deficiency of alignment between driver perception and safe behaviors has implications for the design of future studies and systems that need to balance satisfaction and safety. These observations are likely to improve driver interaction with automated vehicles.
本研究分析了在高速公路变道时，合作式人机界面设计对驾驶员对自动驾驶系统（ADS）的信任和交互的影响。虽然驾驶员对 ADS 的不当信任会影响他们对系统的行为、检测系统性能不足的能力以及对周围交通干扰的感知，但他们参与自动化过程可以提高他们对系统和交通状况的理解，这对于安全实践自动化驾驶是必要的。40 名驾驶员在驾驶模拟器中进行条件自动驾驶实践，并在双车道高速公路的主车道上遇到交通拥堵。提出了四种 ADS 设计来绕过拥堵。ADS-1 检测到拥堵并相应地同步速度。ADS-2 请求驾驶员恢复手动控制并超越拥堵。ADS-3 请求驾驶员按下按钮让系统自动超越拥堵。ADS-4 在通知驾驶员后自动超越拥堵，而驾驶员可以在 6 秒内按下按钮取消。在所有这些情况下，驾驶员干预都是可选的。尽管驾驶员更喜欢和信任 ADS-1 和 ADS-2 而不是 ADS-3 和 ADS-4，但结果表明，在 ADS-3 和 ADS-4 下，驾驶性能和系统使用有显著提高。与其他系统相比，使用 ADS-3 提高了驾驶员的参与度，并减少了对控制转移的需求。然而，主体车辆和相邻车辆之间的时间间隔表明，在 ADS-3 和 ADS-4 下，车道变化比 ADS-1 和 ADS-2 更关键。这种驾驶员感知和安全行为之间的不匹配对未来研究和系统的设计具有重要意义，需要在满意度和安全性之间取得平衡。这些观察结果可能会改善驾驶员与自动驾驶车辆的交互。
