Objective: The aim of this study is to show that by calculating the optimal take-over request lead time (TORlt) using the stabilization time, the stabilization time could be fully utilized in research on take-over and reflect the personal driving tendency that variables in the previous study could not perform.
Background: In conditional automated driving, the driver is placed in an out-of-theloop situation. In order to overcome this problem, it is necessary to take-over request, but there is no official definition of optimal TORlt. In previous studies, they suggested various optimal TORlts, but there is the common limitation that they did not reflect personal driving tendencies. The tendency of personal driving is well observed in longitudinal/lateral control during manual driving, and it is necessary to analyze the stabilization after take-over to see if the driving performance has reached the manual driving level.
Method: A total of 46 subjects participated in the study. This study described a take-over situation due to the sudden appearance of obstacles during automated driving using an OpenDS simulator. After the take-over, the driver performed manual driving until it was determined that the vehicle had stabilized on its own. The driving performance related to vehicle stabilization was measured as the standard deviation of speed and steering wheel angle (SWA). Using these, individual stabilization standards and the stabilization time were measured. Four TORlts (3s, 7s, 10s, 15s) were used as independent variables, and one-factor within-subject ANOVA analysis and Fisher's LSD post-hoc analysis were performed using longitudinal/lateral stabilization time and Dual stabilization time as dependent variables.
Results: In lateral stabilization and dual stabilization, there was a significant difference in stabilization time according to the change in TORlts, which was classified into two groups: 3s and more than 7s. And the stabilization time decreased as TORlt increased from 3s to 10s. But when TORlt increased from 10s to 15s, the stabilization time increased again, which was in a U-shape.
Conclusion: The optimal TORlt calculated using the stabilization time reflecting individual stabilization standards is about 10 seconds, which is similar to previous studies.
Application: This means that the stabilization time can sufficiently replace the role of variables used in the previous take-over studies. In addition, since individual stabilization standards were applied and analyzed, there is a sufficient advantage in that it can reflect personal driving tendencies.