The eDREAM dataset consists of the following parts:

  • EEG data: four channels of EEG signals collected using a wireless Muse headband
  • Physiological data: including ECG, GSR and repiration information
  • Eye-tracking data: gaze and eye-related information estimated by a remote eye-tracker
  • Vehicle-based data: measurements of the driving environment and other
  • Metadata: conditions of experiment variables (e.g. the secondary task) at each sampling time
  • Participant-related data: including subjective rating results, answers to secondary tasks, and experiment logs
  • More details are provided by below. For more information regarding the dataset collection (such as participant group, primary task and secondary task implementation, and procedures), please refer to the data collection report. You are also welcome to contact us for inquiries.

    Vehicle-based Data

    Vehicle-based data were recorded by the driving simulator, NADS MiniSim. The data was originally stored in DAQ formats, which were transformed to MATLAB data files (".mat") using the "ndaqtool" provided by NADS. Each recording contains 383 fields and the description could be found in Appendix B of MiniSimUserGuide-v17. Some of the fields are listed below:

    Field NameColsDescriptionUnits/Values
    Frames1MiniSim frame numberIntegers incrementing at 60Hz
    SCC_LogStreams 1Indicates the current characteristic of the road0 = straight, 1 = curved
    2Indicates if the audio of n-back tasks was on* 0 = no audio, 1 = audio was played
    3Indicates if the lead vehicle was braking 0 = no brake, 1 = braking
    SCC_Follow_Info1Identifier of object-1 = none or error, 0 = no own-vehicle
    2Distance to the lead vehicle (LV)Feet
    3Bumper-to-bumper time to LV Seconds
    4Bumper-to-bumper distance to LVIn feet
    6LV velocityFt/s
    7-9X, y, z coordinate of LV
    CFS_Accelerator_Pedal_Position1Accelerator pedal positionNormalized value between 1 and 0.
    CFS_Brake_Pedal_Force1Brake pedal forcePounds
    CFS_Brake_Pedal_Position1Brake pedal positionRadians of actuator movement
    CFS_Steering_Wheel_Angle1Steering wheel angleDegrees
    ET_frame_num1faceLAB eye tracker frame numberIntegers

    * Or the corresponding duration in no-task drives.

    Physiological Data

    The GSR, ECG and respiration data were recorded using the D-Lab software and exported into text files.

    D-Lab recordings contain the following fields:

    Col NameCol. No.VariableFrequencyDescription
    frame num_frm#1MiniSim Frame Number60 HzRelayed from MiniSim
    frame num_log_stream[0]2During Curve or Not60 HzFirst column of SCC_LogStream from MiniSim:0 = during straight route, 1 = during curve
    frame num_log_stream[1]3Audio States* (n-back test)60 HzSecond column of SCC_LogStream from MiniSim: 0 = no audio, 1 = audio was played
    frame num_log_stream[2]4Leading Vehicle Braking States60 HzThird column of SCC_LogStream from MiniSim: 0 = no braking, 1 = braking
    frame num_SCC_Follow_Info[1]5Headway60 HzSecond column of SCC_FollowInfo in MiniSim: distance to the lead vehicle (LV)
    My Physio System_ECG6ECG240 HzElectrical activity of the heart
    My Physio System_GSR7GSR240 HzElectrical characteristics of the skin, unit: uSiemens
    My Physio System_RESP8Respiration240 HzShape change of torso while breathing

    * Or the corresponding duration in no-task drives.

    EEG Data

    EEG signals were captured using a prototype of Muse headband, and recorded using the MuseLab software. Spontaneous electrical activity from four positions (TP9, Fp1, Fp2 and TP10) were collected. The original data was stored in a binary format (".muse"), which could be converted to CSV, mat-file or printed on screen directly using MusePlayer downloaded from http://developer.choosemuse.com/. The detailed description of the Muse recordings can be found from http://developer.choosemuse.com/research-tools/available-data. Some of the fields are listed below:

    MuseIO PathsFreqUnitDescription
    frameNo60Hz-forwarded miniSim frame number
    /muse/eeg220HzuVRaw EEG data collected from position TP9, Fp1, Fp2, TP10.
    /muse/elements/raw_fft0 /muse/elements/raw_fft1 /muse/elements/raw_fft2 /muse/elements/raw_fft310HzdB129 coefficients representing the logarithm of Power Spectral density for frequencies in the range of 0-110Hz.
    /muse/elements/low_freqs_absolute /muse/elements/delta_absolute /muse/elements/theta_absolute /muse/elements/alpha_absolute /muse/elements/beta_absolute /muse/elements/gamma_absolute10HzBelspower of low frequencies (2.5-6.1Hz), delta band (1-4Hz), theta band (4-8Hz), alpha band (7.5-13Hz), beta band (13-30Hz) and gamma band (30-44Hz) at four positions.
    /muse/elements/delta_relative /muse/elements/theta_relative /muse/elements/alpha_relative /muse/elements/beta_relative /muse/elements/gamma_relative10HzBelsThe relative band powers calculated as percentages of linear-scale band powers in each band for four positions.

    Eye-tracking Data

    Eye tracking data was collected using faceLAB 5.0 at a frequency of 60Hz for all fields. A single faceLAB recording consists of five output under the same folder (with naming convention described at the top of this section). Complete description of all output information can be found in the official Output Data Reference Guide. The table below lists some available data from different files:

    File LabelVariableColumn #Description
    Timing--Experiment timing information (not used).
    HeadFrame number1faceLAB frame number
    Head position (x, y, z)2-4Head position vector in the world coordinate system in meters.
    Head rotation (Q0, Q1, Q2, Q3)5-8A unit quaternion representing the rotation of the ehad coordinate system w.r.t. world coordinate system in radians.
    EyeEye closure (of right and left eyes)2-3The fraction of irirs currently covered by the eye lids. 0.0 = widely open, 1.0=completely closed.
    Blinking71 means blink is occurring and 0 otherwise.
    Gaze rotation (of right and left eyes, w.r.t. x and y axis)15, 16, 21, 22The orientation of the eye gaze w.r.t. the world coordinate.
    Face--Location of facial landmarks in the world coordinate
    WorldV2Gaze intersection item (right, left and both eyes)3, 12, 20Label of the world model item the gaze vector intersects with, which could be "Center_Screen", "Left_Screen", "Right_Screen", "DashBoard", or "Nothing"
    Gaze intersection in World Coordinates (x, y, z for right, left and both eyes)4-6, 13-15, 21-23The point of intersection between the gaze vector and the world model item in world coordinate.

    Face Videos

    A GoPro camera recorded videos for participants 7 to 37 from an approximately 27 degrees angle from participant's right-hand-side. The videos are stored in MP4 format with the following properties:

  • Video encoding standards = h264 (High), yuvj420p
  • Resolution = 720x480 (minimum), 29.97 fps
  • Audio encoding standards = AAC-LC, 48000 Hz

  • Metadata

    Like other recordings, a metadata file is available for each drive (i.e. each participant has three metadata files for each cognitive load scenario). It provides task and road states given frame numbers. This information was extracted from the driving simulator (MiniSim's) log and thus the sampling frequency is 60Hz. The fields in this file is listed below:

    Col. No.VariableValues
    1MiniSim Frame No.Integer
    2FaceLab Frame No.Integer
    3GoPro time Float (second. millisecond)
    4Audio state*0 = no n-back audio
    1 = n-back audio was played
    5Lead vehicle state0 = normal
    1 = intense braking
    2 = slowed
    6Road state0 = straight
    1 = curved
    7Task state*0 = not within the n-back sequence range
    1 = letter sequence was played
    2 = letter sequence was played and lead vehicle braked within this sequence.

    * Or the corresponding duration in no-task drives.

    Particitant-related Data

    Subjective questionnaires used for subjective-rating of perceived workload or risk could be found in appendix of the data collection report. Participants' answers were recorded and organized into an Excel spreadsheet ("Particitants.xlsx").

    Miscellaneous information about each participant's experiment were also recorded in other sheets, including participants' answers to the secondary task, the order of drives, and experiment observations.