An fNIRS dataset for driving risk cognition of passengers in highly automated driving scenarios

A total of 20 participants completed this driving simulator experiment. This study complied with the Declaration of Helsinki and was approved by the Institutional Review Board of Tsinghua University, China. Participants were voluntary after the task and informed consent had been explained, and they also were informed that they may terminate this experiment at any time without any type of penalty. This experiment need participants focus on the front scenarios, so few random stimulating sounds are added in each VTD segment. When doing experiment, participant sits in this driving simulator and looks at the front screen. When he feels dangerous or hears a stimulating sound, he is asked to press the keyboard.This dataset is collected using a human factor signal acquisition system, and it includes a hardware-in-loop equipment, some peripheral equipments, and a host computer. This hardware-in-loop equipment is a real-time system(Redhawk system, and runs Virtual Test Drive (VTD) software; The host computer is a window system and runs Matlab/Simulink, Python module and Oxysoft software. The functions of Matlab/Simulink include the communication between hardware-in-loop equipment and host computer by using User Datagram Protocol (UDP) and the communication between host computer and Oxysoft software. Python module will record the time when they press a keyboard. Oxysoft software and blood oxygen monitoring device work together for acquiring cerebral oxy-haemoglobin change ΔHbO and deoxy-haemoglobin change ΔHbR of passengers. When Matlab/Simulink receives the data about position, velocity and acceleration of ego vehicle and target vehicle (or pedestrian), the raw data of ΔHbO and ΔHbR and participant status, and Matlab/Simulink will joint those data with current time, then compose a new data.When all participants finish this experiment, the new data in same highly automated driving scenario is extracted based on time, and those data is considered as a group. The data of each participant in a highly automated driving scenario is considered as a sample data, and the sample data contains the information of position, velocity acceleration about ego vehicle and target vehicle (or pedestrian) and the raw data of ΔHbO and ΔHbR. Besides, the sample data also contains a kinetic energy field and a flag bit, the kinetic energy field may indicate a danger degree of driving scenario, and it involves relative longitudinal distance and the speed of target vehicle (or pedestrian); The flag bit indicates that whether or not the period in a sample time contains a stimulating sound, if it is true, then the period contains a stimulating sound, or not.

本驾驶模拟器实验共有20名受试者参与完成。本研究符合《赫尔辛基宣言》相关伦理要求，并已通过中国清华大学机构审查委员会（Institutional Review Board）的审批。受试者均为自愿参与，实验前已向其充分说明知情同意事项，同时告知其可随时终止实验且不会受到任何处罚。本实验要求受试者专注于前方驾驶场景，因此在每个Virtual Test Drive（VTD）片段中会插入少量随机刺激性音效。实验过程中，受试者坐在驾驶模拟器座舱内，注视前方显示屏；当感知到危险或听到刺激性音效时，需按下键盘作出响应。本数据集通过人机工效信号采集系统采集得到，该系统包含硬件在环（Hardware-in-loop）设备、若干外围设备以及一台上位机。其中，硬件在环设备为实时系统（Redhawk系统），搭载VTD软件；上位机运行Windows系统，同时部署Matlab/Simulink、Python模块及Oxysoft软件。Matlab/Simulink的功能包括：通过用户数据报协议（User Datagram Protocol, UDP）实现硬件在环设备与上位机之间的通信，以及上位机与Oxysoft软件之间的通信。Python模块用于记录受试者按下键盘的时刻。Oxysoft软件与血氧监测设备协同工作，用于采集受试者的脑血氧合血红蛋白变化量ΔHbO与脱氧血红蛋白变化量ΔHbR。当Matlab/Simulink获取自车、目标车辆（或行人）的位置、速度及加速度数据，ΔHbO、ΔHbR原始数据以及受试者状态数据后，将结合当前时间戳对上述数据进行整合，生成新的数据集条目。待所有受试者完成实验后，基于时间戳从同一高度自动驾驶场景的整合数据中提取对应片段，将同一场景下的所有片段划分为一个数据集组。单一场景下每名受试者的对应数据即为一条样本数据，其包含自车、目标车辆（或行人）的位置、速度与加速度信息，以及ΔHbO、ΔHbR原始数据。此外，样本数据还包含动能场与标记位：动能场用于表征驾驶场景的危险程度，其参数涵盖目标车辆（或行人）的相对纵向距离与运动速度；标记位用于标识样本时段内是否存在刺激性音效，若标记为真，则该时段包含刺激性音效，反之则无。