Towards Multimodal Driver’s Stress Detection - Digital Signal Processing for In-Vehicle Systems and Safety - page 6

Table 1.1 UTDrive data collection protocol

Secondary tasks
A

Part B C

Route1 1 Lane changing Common tasks (radio,

AC etc.)

Sign reading

2 Cell phone dialog Cell phone dialog Conversation

3 Common tasks Sign reading Spontaneous

4 Conversation Spontaneous Cell phone dialog

Route2 1 Sign reading Lane changing Common tasks (radio,

AC etc.)

2 Cell phone dialog Cell phone dialog Conversation

3 Common tasks (radio,

AC etc.)

Sign reading Lane changing

4 Spontaneous Conversation Sign reading

Session Route Task

1 1 Just drive

1 Secondary tasks A

2 Secondary tasks A

2 Just drive

2 1 Just drive

1 Just drive

2 Secondary tasks B

2 Secondary tasks C

3 2 Secondary tasks C

1 Secondary tasks C

2 Just drive

2 Just drive

to have six labels to denote different maneuvers of the driver. A color-coded driving
timeline is developed to observe aligned task and event transcriptions to obtain more
insight into the data as well as to see the overlapping sections between tasks and
events. A detailed explanation is given in [ 13 ] for transcription labels and color-coded
driving timeline.
It should be noted that cell phone dialog includes different types of distractions:
manual (dialing and holding), cognitive (interaction and processing), and auditory
(listening). Therefore, the segment of the road containing the cell phone dialog can
be considered as the highest possibility of observing high levels of distraction and
divided attention. Although the cell phone in the car interfaces via a bluetooth
device and the manual tasks from the driver minimized, the initial dialing might
cause momentary distraction.

1.3 Stress Detection Using Speech Signal

This section focuses on the stress assessment from the driver's speech. First, it
should be noted that the real level of stress in the driver caused by the cognitive load
is not known. To define stress levels in the speech segments, we apply a cause-type