Drivers Drowsiness Database: A collection of physiological signals during the use of a driving simulator (DD-Database)

The DD-Database contains the physiological signals of 4 EEG channels, 2 EOG channels and 1 ECG channel as well as annotation files corresponding to 10 healthy volunteers between 20 and 50 years old. The signals were collected during the use of a driving simulator under a protocol designed to induce drowsiness. The annotation files have the time marks of subject´s drowsiness events. The experiment was carried out in 2 trials of 2 hours each, so the database contains 40 hours of information. This database is a contribution for the development and evaluation of algorithms and/or systems for detecting drowsiness in drivers. Data acquired during the use of a driving simulator and publicly available, are scarce.

Methods
Data were collected from 10 healthy volunteers between 20 and 50 years old during the use of a driving simulator. All volunteers provided written informed consent. For all subjects, approximately 4 hours of physiological signals were collected.

The experiment itself consisted of acquiring physiological signals during the use of a driving simulator in a dark and quiet environment while presenting a night driving scenario on a straight route. The day before the test, the subjects were instructed to be partially deprived of sleep, attend after lunch at the agreed time and to not take medications, or alcoholic or energy drinks. All these actions pointed to induced drowsiness. The test lasts 2 hours and it was carried out twice for each volunteer on a different day. Additionally, the volunteers were instructed to press an event button when they felt drowsy during the test.

The data are de-identified and free of personal health information (PHI). The volunteer number (01 to 10) and the trial number of the same volunteer (1 or 2) are randomly assigned, both trails always remain together. The information of the gender and the channel is kept.

The physiological signals were acquired at a sample frequency of 128 Hz with a 16-bit resolution A/D converter. EEG and EOG were configured as unipolar channels. For ECG a bipolar channel in DIII position was acquired. An AKONIC Neurotrace-Mini PC polysomnography equipment was used for the signal acquisition. The Mini-PC model has a sampling frequency of 128 Hz for all signals, 32 channels plus the system reference and 7 ground inputs filters