Hwang Jiwoo, Choi Woohyeok, Lee Jungmin, Kim Woojoo, Rhim Jungwook, Kim Auk
Kangwon National University, Chuncheon, 24341, South Korea.
Sci Data. 2025 Mar 31;12(1):539. doi: 10.1038/s41597-025-04781-8.
Automated driving systems enable drivers to perform various non-driving tasks, which has led to concerns regarding driver distraction during automated driving. These concerns have spurred numerous studies investigating driver performance of fallback to driving (i.e., takeover). However, publicly available datasets that present takeover performance data are insufficient. The lack of datasets limits advancements in developing safe automated driving systems. This study introduces TD2D, a dataset collected from 50 drivers with balanced gender representation and diverse age groups in an L2 automated driving simulator. The dataset comprises 500 cases including takeover performance, workload, physiological, and ocular data collected across 10 secondary task conditions: (1) no secondary tasks, (2) three visual tasks, and (3) six auditory tasks. We anticipate that this dataset will contribute significantly to the advancement of automated driving systems.
自动驾驶系统使驾驶员能够执行各种非驾驶任务，这引发了人们对自动驾驶过程中驾驶员注意力分散的担忧。这些担忧促使众多研究对驾驶员接管驾驶（即接管）的表现进行调查。然而，提供接管性能数据的公开可用数据集并不充足。数据集的缺乏限制了安全自动驾驶系统开发的进展。本研究引入了TD2D，这是一个从50名驾驶员在L2自动驾驶模拟器中收集的数据集，这些驾驶员具有平衡的性别代表性和不同的年龄组。该数据集包括500个案例，涵盖了在10种次要任务条件下收集的接管性能、工作量、生理和眼部数据：（1）无次要任务，（2）三种视觉任务，以及（3）六种听觉任务。我们预计该数据集将对自动驾驶系统的发展做出重大贡献。
