Re-use of FOT Data to Assess the Frequency and
Impact of Distracted Driving
Andreas Landau
Dr. Barbara Metz
Dr. Volker Hargutt
Alexandra Neukum
Naturalistic Driving Research Symposium
Blacksburg, Virginia, 26th of August, 2014
2
Analysis methods
Data base (Objective data,
video,…)
Driver Surrounding
Culture Vehicles
… Analysis methods
Data base
• EU-project „euroFOT“ to evaluate driver assistance
systems
• Data from 115 drivers provided with a well-equipped car for
three months
• Three variants of navigation (without navigation
system/build in/mobile)
• Statement of the drivers if a trip is familiar or not at the
beginning of a trip
• No other restrictions => Field operational test (FOT)
=> Naturalistic driving study (NDS) for analysis of
distracted driving
3
Data base
Results after
preprocessing steps
Overall
Number of Participants 115
Number of Participants with complete data set
104
Number of Trips 39 703
Observed Kilometers 1 013 262
Observed Hours 15 129
4
• A few hundred variables with objective data (speed,
THW, use of systems, driver-vehicle-interaction,…)
• Video data from
nearly all trips
(4 cameras)
• Coding of 348 trips
with 20000 km and
256 hours overall for
analyzing distracted
driving.
Data base
5
Procedure
6
Objective Data
Frequency of distracted driving for each driver
Video coded data
Analysis of distracted driving for each
driver
Results showing the benefit of NDS
Frequency of distracted driving for each driver
• Objective data: Talking to a hands-free telephone is the most
frequent task.
• Video-coded data: Vehicle and mobile-interaction are the most
frequent tasks.
Frequency of distracted driving
7
results from CAN-data
talk
ing h
and-f
ree p
hone
centr
al button
buttons a
t ste
eri
ng w
heel
mid
dle
console
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
20%
pro
port
ion tim
e [%
]
results from video coding
vehic
le r
ela
ted
h
andl. m
obile p
hone
eat / dri
nk
talk
ing m
obile p
hone
sm
okin
g
bodily r
ela
ted
oth
er
technic
al devic
es
searc
hin
g
read / w
rite
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
20%
pro
port
ion o
f codedf tim
e [%
]
• Distracting activities are mostly located in the area of the middle
console and directly in front of the driver.
• Drivers mostly either use no hand or the right hand for the
secondary task.
Frequency of distracted driving
8
Where ?
How ?
• Choosing an appropriate analysis method to find reliable
effects of distracted driving.
=> Ensure comparability
• Different type of secondary tasks (navigation vs. radio)
• Secondary tasks have different length (telephone vs. pressing
a button).
Impact of distracted driving
Secondary task
Before Beginning End After During
A trip
Telephoning No Telephoning
Telephoning Telephoning
Time-based approach Sequence-based-approach
9
Sequence-based approach for outgoing calls (N=92 drivers)
• A significant increase of the distance for all road classes
can be found.
• The results for incoming calls are similar.
Motorway
no phoning
phoning
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
TH
W [
se
c]
Rural
no phoning
phoning
0.0
0.4
0.8
1.2
1.6
2.0
2.4
2.8
TH
W [
se
c]
Urban
no phoning0
1
2
3
4
5
10
Impact of distracted driving – phoning
Time-based-approach for handling turn-and-press-controller
• Significant increasing of THW during handling.
• For speed similar results can be found in rural and urban
areas.
Impact of distracted driving –
turn-and-press-controller
11
Motorway Rural Urban0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0m
(TH
W)
[Sec]
before
during
after
• Significant less lane changes during handling.
• During handling significant less time on crossings.
12
Impact of distracted driving –
turn-and-press-controller
motorway rural urban-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
La
ng
e c
ha
ng
es p
er
min
ute
[N
/min
]
before
during
after
rural urban0%
2%
4%
6%
8%
10%
12%
14%
Pro
port
ion o
f tim
e o
n c
rossin
gs [
%]
before
during
after
Secondary task
Lane Change Lane Change
Number [N] Time [min]
Secondary task
Proportions of time
Crossing No Crossing
Conclusion
• Objective data (> 100 drivers, ~1 000 000 kilometers) and video-
coded data (>100 drivers, ~ 20000 kilometers) were analyzed.
• Phoning and vehicle interaction are the most frequent tasks.
• Two approaches were used for analyzing impact of distracted driving
•
• Adaptation of driving parameters during secondary tasks as well as
hints for a situation-aware behavior can be found.
• The benefit of NDS depends on variety of influences and the choice
and development of appropriate analysis methods.
13
Thank you! Würzburger Institut for Traffic Sciences GmbH (WIVW)
Robert-Bosch-Straße 4
97209 Veitshoechheim
Tel.: +49-(0)931-78009102
Fax: +49-(0)931-78009150
e-mail: [email protected]
Andreas Landau
Dr. Barbara Metz
Dr. Volker Hargutt
Alexandra Neukum
