Citation

Arakawa T, Hibi R, Fujishiro T. Transp. Res. A Policy Pract. 2019; 124: 587-610.

Copyright

(Copyright © 2019, Elsevier Publishing)

DOI

10.1016/j.tra.2018.05.003

PMID

unavailable

Abstract

Recently, there has been an increasing interest in the development of autonomous vehicles. However, some of the challenges that are associated with autonomous vehicles are yet to be resolved. Here, we investigate the user dependence on autonomous control in autopilot vehicles. Further, we verify whether drivers can control the vehicles in an appropriate manner after a system failure based on the range of biometric data. The participants in this study experienced three scenarios in a driving simulator: manual-driving, autonomous-driving, and system-failure scenarios, which forced the participants to resume manual control. The data obtained from salivary amylase depict that drivers tend to be anxious while performing the transition from the manual-driving scenario to the autonomous-driving scenario; however, all drivers have to be accustomed with the system in order to drive using a driving simulator. Thus, the difference of anxiety between various drivers is observed to be suppressed. The seat pressure data illustrate that an area having a value equal to 95% of the probability-deviation ellipse during the second day of the manual-driving and autonomous-driving scenarios is observed to have the largest value from among each driving scenario, which seems to be caused due to the driver's concentration or fatigue. Further, the systolic blood pressures of all the drivers increased with time. However, in the autonomous-driving scenario, the average relative systolic blood pressure is, on the whole, higher than that in the manual-driving scenario. It is suggested that the drivers had never previously used an autonomous-driving system and that he/she may be uncomfortable or uneasy to autonomously control the brake, accelerator, and steering wheel. Additionally, after a system failure during the manual-driving scenario, the behavior is observed to be similar to that in the initial manual-driving scenario, and the totally average relative systolic blood pressure is higher than that during the initial manual-driving scenario. This indicates that the drivers' mental workload is relatively low during the autonomous-driving scenario because they do not experience any stress from driving. However, the drivers' systolic blood pressure increased because of the transition from autonomous driving to manual control and because of the mental workload to control the vehicle on their own just after using an autonomous-driving system. From the viewpoint of the brain activity in the left frontal lobe, the data indicates that the drivers' cognition level during autonomous driving is lower than that during manual driving and that the declining tendency of the average relative hemoglobin concentration is remarkable during the manual-driving scenario after encountering a system failure. This is because the driver feels that he/she may commit mistakes during manual driving. Additionally, if he/she is driving an autonomous vehicle and if the autonomous-driving system failed, he/she does not need to pay attention to the surrounding subjects and control the vehicle. The eye-gaze data indicate that "mind distraction" occurred in the participants while resuming control after a system failure because their brain activity at this instance was relatively low. Our results indicate that drivers who depend on autonomous control systems experience stress upon switching to manual control after a system failure.

Language: en

Keywords

Autonomous vehicle; Dependency; Driving simulator; Human factors; System failure