Content uploaded by Shengbo Eben Li
Author content
All content in this area was uploaded by Shengbo Eben Li on Dec 18, 2018
Content may be subject to copyright.
International Journal of Automotive Technology, Vol. 16, No. 5, pp. 739−750 (2015)
DOI 10.1007/s12239−015−0075−5
Copyright © 2015 KSAE/ 086−03
pISSN 1229−9138/ eISSN 1976−3832
739
FIELD OPERATIONAL TEST OF ADVANCED DRIVER ASSISTANCE
SYSTEMS IN TYPICAL CHINESE ROAD CONDITIONS: THE INFLUENCE
OF DRIVER GENDER, AGE AND AGGRESSION
G. LI, S. EBEN LI
*
and B. CHENG
State Key Laboratory of Automotive Safety and Energy, Department of Automotive Engineering,
Tsinghua University, Beijing 100084, China
(Received 27 October 2014; Revised 29 January 2015; Accepted 11 March 2015)
ABSTRACT−Although various Advanced Driver Assistance Systems (ADASs) have been developed to assist drivers, their
performances and driver acceptances in China have not been well tested and analyzed. This study aims to examine how do
driver gender, age, and aggression affect the performances and driver acceptances of typical ADASs by means of Field
Operational Tests (FOTs), including FCW (Forward Collision Warning), LDW (Lane Departure Warning), and SBZA (Side
Blind Zone Alert). Thirty-three participants were recruited to drive an equipped vehicle on the test route in and around Beijing
City. Vehicle states, environmental information, and driver feedback were recorded by CAN bus, cameras, and post-drive
questionnaires. The test results showed that the alert frequencies of FCWs and LDWs increase in higher speed traffic
scenarios, whereas that of SBZA declines. Driver acceptance rate of SBZA ranks the highest, with FCW ranking the second
and LDW being the last. Driver gender, age, and aggression effects were analyzed in details, showing their relationships with
total alert times, alert times per 100 km, and driver acceptance rate of each system. The findings are helpful for future
development of ADASs for automotive industry.
KEY WORDS : ADASs performance, Driver acceptance, Gender, Age, Driver aggression, Field operational test
1. INTRODUCTION
Rear-end, angle and sideswipe crashes accounted for 32.2%,
22.7% and 9.8% of all road crashes in the US in 2010,
respectively (NHTSA, 2012). The numbers were 40.4%,
6.6% and 4.1% for highway crashes in China in the same
year (Ministry of Public Security Traffic Management
Bureau, 2011). To improve road safety, various Advanced
Driver Assistance Systems (ADASs) were developed to
provide drivers specific features, e.g., Forward Collision
Warning (FCW) system, Lane Departure Warning (LDW)
system, and Side Blind Zone Alert (SBZA) system. The
potential functionality of ADAS on reducing road crashes
has been recognized by both automotive industry and
academia. In the NHTSA research priority plan for 2011 ~
2013 (NHTSA, 2011), FCW is in the list of Priority
Projects, and the other two systems (LDW and SBZA) are
included in the list of Other Significant Projects.
From an operational point of view, the development of
such systems clearly departs from traditional driving tasks
at all times (Piao and McDonald, 2008). Instead, ADASs
could help or replace drivers on some decisions and actions.
This makes it possible to eliminate many pre-accident
human errors and achieve more benefits on driving safety
and fuel economy than before.
FCW is developed to alert a driver to avoid or mitigate
the imminent collision with an obstacle ahead of the
subject vehicle. Based on laser or radar technologies, FCW
can measure inter-vehicle distance, angular position, and
relative speed with the target vehicle ahead, and alert
drivers in various visual/audio/haptic ways (Jeong and
Green, 2012). Some enhanced systems, to further assist
drivers, have additional functions to support and substitute
drivers to control the brake (Floudas et al., 2004; Shaout et
al., 2011; Oikawa et al., 2014). Wandering out of the
current lane may result in severe collisions with vehicles in
adjacent lanes. To reduce this type of crashes, LDW is
developed to help avoid departure dangers. Embedded with
a camera to recognize lane markers, LDW is activated to
present warnings when an unintended lane departure is
recognized. Steering wheel vibration and/or slight auto-
matic correction may also be provided when needed in
some systems (Suzuki and Jansson, 2003). SBZA provides
another kind of lateral movement assistance to help drivers
avoid lane change crashes. Typically using ultrasonic radar
technology to detect vehicles in side blind zones, SBZA
alerts drivers to prevent crashes primarily caused by “did
not see other vehicles” when planning a lane change
maneuver. Besides the most common visual and audio
alerts, assistance in steering has also been available in some
*Corresponding author. e-mail: lisb04@gmail.com
740 G. LI, S. EBEN LI and B. CHENG
products (Shaout et al., 2011).
It has been found that crashes can be effectively avoided
or mitigated if the interference strategies of ADASs are
provided correctly and properly. For example, according to
an investigation conducted by National Transportation
Safety Board, 60% of rear-end crashes could be avoided if
an FCW alarm was given 0.5 second ahead, and around
90% of rear-end accidents could be avoided if the alarm
could be given 1.5 seconds ahead (National Transportation
Safety Board, 2001). Lee and Jeong (2014) proposed a
method to generate warnings at different times according to
the driving peopensity determinded by three metrics,
including predicted time headway, required deceleration,
and resultant acceleration. Properly activated warnings to
avoid crashes were verified on both test groud and public
roads. It is estimated by some case studies that ADASs can
prevent up to 40% of traffic accidents, varying across
system types and accident scenarios (Zador et al., 2000;
Jagtman et al., 2001; Golias et al., 2002). Although the
original purpose of ADASs is to affect traffic safety
positively, negative effects have been found as well (Lindgren
and Chen, 2006; Dragutinovic et al., 2005; Saad, 2004;
Brookhuis et al., 2001). Driver’s reaction delay and false
alert nuisance are reported most (Brookhuis et al., 2001).
Accordingly, driver trust and acceptance of ADASs vary
across drivers depending on both their performance and
driver characteristics (Lindgren and Chen, 2006).
Many Field Operational Tests (FOTs) have been con-
ducted to evaluate ADASs performances and to examine
their driver acceptance. Alkim et al. (2007) carried out an
FOT to perform an objective assessment on ACC (Adaptive
Cruise Control) and LDW, and found a promising result
that driving with ACC and LDW could improve traffic
safety with approximately 8% reduction on time headway
and 3% improvement on fuel economy. Adell et al. (2011)
examined the effects of the driver assistance system on
keeping safe distance and speed. The findings showed both
positive and negative effects in terms of safety concerns.
Considering driver gender and age, Najm et al. (2006)
conducted a test to characterize the performance and to
determine driver acceptance of an FCW. Results showed
that older driver group were more willing to rent or
purchase an FCW-equipped vehicle than younger groups.
Ervin et al. (2005) conducted a test to examine the
suitability of an FCW from the perspectives of both driving
safety and driver acceptance. Results indicated that the
acceptance of FCW was mixed due to false alarms and was
not found to be significantly related to FCW alert rate.
Similar tests were carried out on other types of FCW,
SBZA, and LDW (LeBlanc et al., 2006; Sayer et al., 2011).
Besides FOTs, Lee and Peng (2005) developed an alter-
native method to identify ‘threatening’ and ‘safe’ data sets
and used it to evaluate the performance of five published
collision warning systems by simulations.
Driver characteristics can significantly affect the ADAS
performance and driver acceptance. It has been found that
driver gender and age made independent significant contri-
butions to traffic accident involvement (Reason et al.,
1990), thus leading to diversities on ADASs performance
and subjective acceptance across gender and age groups
(Najm et al., 2006; LeBlanc et al., 2006; Sayer et al.,
2011). Besides gender and age, driver aggression is another
important factor needs to be considered (Xie and Parker,
2002). Self-report questionnaire is a most popular way to
investigate drivers’ aggression. Buss and Perry (1992) pro-
posed an aggression questionnaire to measure people’s
physical aggression, verbal aggression, anger, and hostility.
Correlation analysis results showed that anger is the bridge
between each two of the other three scales. Specifically for
driving anger scale, Deffenbacher et al. (1994) measured
how much a driver would be irritated by driving-related
situations. Since China has a booming automotive industry,
and Chinese drivers behave differently with drivers abroad
(Zhang et al., 2006; Li et al., 2010), it is important to know
how ADASs work in typical Chinese road conditions. To
date, although many studies have been reported in China,
few FOT attempts in naturalistic traffic flow have been
made.
The purpose of this paper is to conduct an FOT in typical
Chinese road conditions to examine how do driver gender,
age, and aggression affect the performances and driver
acceptances of three commonly used ADASs, including
FCW, LDW, and SBZA. The remainder of this paper is
structured as follows: Section 2 describes details of the
experiments, including data collection system, test route
design, and participants’ information; Section 3 presents
the affection of driver gender, age, and aggression on
ADASs performances and driver acceptances; Section 4
discusses the test results and summarizes the flaws needed
to be improved for future ADAS development; Section 5
concludes this paper.
2. METHODOLOGY
2.1. ADASs Equipped in the Test Vehicle
The test vehicle used in this study is a passenger car with a
4.6L internal combustion engine and a 5-speed automatic
transmission. Equipped ADASs include FCW, SBZA, and
LDW. Both visual and auditory warnings are provided in
all three systems. The FCW is automatically functional to
use radars to detect target vehicles when the subject vehicle
speed exceeds 32 km/h (20 mph). When a vehicle is
detected, a green-car-icon will be displayed on a screen
mounted on the upright side out of the dashboard. Caution-
ary alerts, indicated by a yellow-car-icon, will be visually
presented to the driver when the following distance is
closer. Consisted of a red-collision-icon and a buzzing
sound, crash imminent alerts will be presented when the
following distance is too close. The LDW is activated
when the speed is higher than 56 km/h (35 mph) and at
least one side of lane markers is detected. The camera used
to detect lane markers is mounted near the rearview mirror.
FIELD OPERATIONAL TEST OF ADVANCED DRIVER ASSISTANCE SYSTEMS IN TYPICAL CHINESE 741
When lane marker is detected, a green icon will be dis-
played on the dashboard. The color will change to red and
flash with low buzzing sound when the driver departs the
lane without using turning signal. The SBZA uses radars to
detect vehicles 5 m behind rearview mirrors in adjacent
lanes. An icon will be displayed in the rearview mirror
when a vehicle in the side blind zone is detected. Mean-
while, if a driver intends to move to the next lane without
using turning signal, the icon will flash with a displayed
arrow.
2.2. Test Route
Considering the affection of traffic situations on different
road types, the test route in and around Beijing (between
Tsinghua University in Beijing and Xianghe in Hebei
province) was selected for experiments. As shown in
Figure 1, city roads (a), city expressways (b), and inter-city
highways (c) were included, totally about 225 km. Total
distance of section a was about 23 km and the speed limits
varied from 30 to 60 km/h. The fourth ring-road, marked as
b in Figure 1, was selected as city expressways. The total
distance was about 56 km and the speed limit was 80 km/h.
A section of Beijing-Harbin highway, marked as c in the
figure, was selected as inter-city highways. The total distance
was about146 km and the speed limit was 120 km/h.
Participants were given route guidance instructions verbally
by an experiment assistant present in the vehicle all the
time in the experiment. This was provided to ensure all
drivers received the same instructions according to a pre-
determined script. All participants drove the same instru-
mented vehicle throughout the study.
2.3. Data Collection System
Six cameras were installed in the test vehicle to record the
road situations (channel A: front road image, channel E: left
blind zone, and channel F: right blind zone), driver status
and operations (channel C: face image, channel D: foot
image), and warning system status (channel B: FCW warn-
ing status). This image collection system saved images into
a hard-disk recorder at 25 Hz. Another camera (camera G)
was equipped to record the front road images for data
synchronization. See Figure 2 for the architecture of the
data collection system and Figure 3 for an example of the
collected images. Recorded CAN bus data include vari-
ables about vehicle state, driver operation and warning
system alert timing. Figure 4 shows a sample of the
collected CAN bus data. This data acquisition system saves
both CAN bus data and channel G images at 10 Hz into a
laptop. To synchronize data, channel A images were used to
match the front road images captured by channel G.
2.4. Participants
The following criteria were required for participants:
(1) Driver age must be between 20 and 65.
Table 1. Participants information.
Participant
number
Age Driving
experience
Mean SD Mean SD
All 33 43.9 10.8 14.2 7.7
Male 22 45.0 11.1 16.0 8.2
Female 11 41.7 10.2 10.4 5.0
Figure 1. Selected test route in and around Beijing.
Figure 2. Data collection system architecture.
742 G. LI, S. EBEN LI and B. CHENG
(2) Driving experience should be no less than 2 years.
(3) The driver must have his/her own car.
(4) The driver should not have previous driving experience
with ADASs.
Note that the last criterion was required to avoid inter-
ference on driver behaviors and subjective ratings from his/
her previous driving experience.
Finally, thirty-three participants (22 male and 11 female)
were recruited to take part in this study, with an accumula-
tion of 7,500 km of driving. Their age ranged from 28 to 65
years old (Mean = 43.9, SD = 10.8). On average, they had
14.2 years driving experience, ranging from 3 to 33 years
(SD = 7.7). Detailed information about the male and female
participants can be found in Table 1.
2.5. Procedure
A set of training was provided prior to starting the FOT.
They were given a verbal explanation on what they need to
do and what would happen during the experiment. To get
used to the systems, a video on how the ADASs work was
exposed to the participants as well to help them better
understand their capabilities and limitations. The assistant
would check the working status of the vehicle, data acqui-
sition system and image collection system before starting
the experiment. It would roughly take about 80 minutes to
drive from the start point to the rest point. Drivers could
take a break at the rest point, and then it would take another
80 minutes to drive back to the end point.
After the experiment, subjective data were collected
through a simple questionnaire as well as an interactive
debriefing that was done to collect drivers’ advice on
systems improvement. All the three statements listed in the
questionnaire were I am satisfied with the FCW/SBZA/
LDW, specifically. Using Likert Scale (Joy, 2007), subjects
were asked to rate a value from 1 to 5 indicating how much
they agree with each of the statements, with 1 representing
Strongly disagree and 5 representing Strongly agree.
Another questionnaire was also presented to drivers to
investigate their driving aggression level. This questionnaire
was mainly developed from two previous studies (Buss and
Perry, 1992; Deffenbacher et al., 1994), and was split into
three sections: driver characteristics, driving environments,
and driver preferences. Subjects were required to rate each
item shown in Table 2 in terms of how characteristic they
were of the subject himself/herself (for driver characteri-
stics and driver preferences questions) or how likely would
a specific traffic situation irritate the driver to be angry (for
driving environment questions). Seven-level Likert Scales
were adopted. The sum of all the rating scores came up to a
driver’s aggression score. The higher the number was, the
more aggressive the driver would probably be. Cronbach’s
alpha was adopted to evaluate the internal consistency of
the scale scores (Cronbach, 1951). The alpha values for
driver characteristics, driving environments, and driving
preferences scales were 0.76, 0.88, and 0.83, respectively.
See Table 3.
Figure 3. Example of the images and warning system data.
Figure 4. Example of the CAN bus data.
FIELD OPERATIONAL TEST OF ADVANCED DRIVER ASSISTANCE SYSTEMS IN TYPICAL CHINESE 743
3. RESULTS AND ANALYSIS
Thirty-three participants drove a total of about 7,500 km
during the FOTs. The data collection system did not work
well in two of the tests. Besides, another driver did not take
the aggression questionnaire seriously, filling 67% of the
questions with the most negative answer which did not
match his driving image record at all, so that his data was
eliminated in the analysis concerning driver aggression.
The analysis of variance (F-test) is used when the
normality and homoscadicity requirements are satisfied.
Otherwise, a nonparametric test (Kruskal-Wallis one-way
analysis of variance) is used to test whether the results
originate from the same distribution.
3.1. Overall ADASs Performance and Driver Acceptance
Concerning the overall performance of ADASs, three
drivers did not receive any FCW alert during the driving
back and forth. The mean alert times per driver for FCW,
SBZA, and LDW were 9.8 (SD = 10.8), 37.1 (SD=21.0),
and 92.2 (SD=62.5), respectively.
To further analyze ADASs alert frequency in various
traffic situations, alert times per 100 km for each driver
(referred to as ATPK) was used as an evaluation index. As
shown in Figure 5, almost no FCW alerts occurred on city
roads. That is because the FCW would only be triggered to
be functioned when vehicle speed was higher than 32 km/
h. However, heavy traffics on selected city roads forced
vehicle speed to be under 32 km/h most of the time. For
city expressways and inter-city highways, analysis on speed
distributions indicated that driving speed ranged from 50 to
75 km/h and from 105 to 120 km/h individually, aligning
with the posted speed limits. Thus, indicated by Figure 5,
higher speed would contribute to more FCWs and LDWs,
Table 2. Driver aggression questionnaire.
Questions
Driver characteristics
1 My friends say that I’m somewhat argumentative.
2 I tell my friends openly when I disagree with them.
3 I can’t help getting into arguments when people dis-
agree with me.
4 I get into fights a little more than the average per-
son.
5 I have trouble controlling my temper.
6 I have become so mad that I have broken things.
Driving environments
7 Someone starts moving late, when the traffic light
signalizes green.
8 Cyclist is cycling in the middle of the lane and
slowing the traffic.
9 Someone drives very slowly in the fast left lane and
you want to overtake.
10 Someone is pushing on the back of your car (e.g.,
wants to force the release of the lane).
11 Someone cuts in front of you for the parking place
you are waiting for.
12 Someone drives between lanes and blocks your
way.
13 Someone is reversing in front of you without look-
ing back.
14 Someone in the opposite direction does not dim his/
her lights.
15 Someone increasing the vehicle speed when you are
trying to overtake hem/her.
16 Someone made inappropriate gestures to you
because of your style of driving.
Driver preferences
17 I like driving faster than others.
18 I start earlier than others when the traffic light turns
to green.
19 I brake harder than others.
20 I change lanes in a shorter time than others.
21 I follow a lead vehicle closer than others.
22 I like to drive as I want for sensation seeking (e.g.,
drag racing).
23 I think my driving skill is better than the average
level.
24 I think I am an aggressive driver.
Table 3. Means, standard deviations, and alpha reliability
coefficients for three driver aggression scales.
Number
of items Mean SD α
Driver characteristics 6 1.68 0.50 0.76
Driving environments 10 3.05 0.97 0.88
Driver preferences 8 2.39 0.93 0.83
Table 4. Significance test results of ATPK.
City
roads
City
expressways
Inter-city
highways p
FCW ATPK 0.0 2.7 5.3 < 0.001
SBZA ATPK 44.9 29.0 5.6 < 0.001
LDW ATPK 14.5 29.6 39.1 < 0.001
Figure 5. Overall ATPK in different traffic situations.
744 G. LI, S. EBEN LI and B. CHENG
but less SBZAs. Statistical significances were found for all
systems. See Table 4. The mean differences between each
two traffic situations were also proved to be significant ( p
< 0.05) in pairwise comparisons tests.
The mean subjective ratings on the SBZA were 3.9
(SD= 0.6), compared to 3.2 (SD = 2.3) on the FCW and 3.2
(SD= 1.0) on the LDW. Define driver acceptance rate here
as the percentage that a driver rates a system as 4 or 5. As
shown in Figure 6, drivers never scored FCW or SBZA as
1. Although LDWs were received far more frequently than
any of the other two kinds of warnings, it got the lowest
acceptance rate (32.3%). The FCW and SBZA got an
acceptance rate of 50.0% and 67.7%, respectively.
3.2. Gender
The alert times per driver of each warning system were
shown in Figure 7. On average, male drivers triggered
162% more FCWs than female drivers, but 25% fewer
LDWs. The more FCW alert times for male drivers (p=
0.065) is probably because male drivers tend to drive more
aggressively than female drivers in hazardous situations
(e.g., driving through heavy traffic) and in dangerous ways
(e.g., rapidly approaching and closely following). SBZAs
did not vary much between genders (Male: 31.8, Female:
33.8).
As illustrated in Figure 8, ADAS ATPK as a function of
driver gender varies across traffic situations. When com-
paring FCW ATPK between males and females on city
expressways, the former showed a higher probability to
trigger FCWs. Statistical significance was found (p=
0.005). The observed mean FCW ATPK for females on city
expressways was 0.4 (SD=0.8) compared to 4.2 (SD=4.0)
for males. Similar trend was shown for FCW ATPK on
inter-city highways. No significant differences were found
for SBZA ATPK and LDW ATPK.
Female drivers viewed each system more positively than
male drivers in general. The acceptance levels male drivers
ascribed to FCW, SBZA, and LDW were at an average of
3.4 (SD= 1.1), 3.7 (SD=0.7), and 3.1 (SD=0.9), respectively,
while female drivers rated them at an average of 3.7 (SD
=1.2), 4.3 (SD=0.8), and 3.5 (SD=1.1). Significance was
found for SBZA (p=0.042). The observed acceptance rate
Figure 6. General driver acceptance of each system.
Figure 7. Alert times per driver as a function of driver
gender.
Figure 8. ATPK in different traffic situations as a function
of driver gender.
Figure 9. Driver acceptance as a function of driver gender.
FIELD OPERATIONAL TEST OF ADVANCED DRIVER ASSISTANCE SYSTEMS IN TYPICAL CHINESE 745
of SBZA for females was 82%, 22% higher than that for
males. Similarly, females rated FCW and LDW 9% and
20% higher than males, respectively. See Figure 9.
3.3. Age
According to their age, subjects were divided into three
groups: young (21 to 35 years old), middle (36 to 50 years
old), and old (51 to 65 years old). The mean age for each
group was 30.3 (SD = 2.9), 41.9 (SD = 3.7), and 57.9 (SD=
2.8), respectively. As indicated in Figure 10, the total alert
times per driver did not vary much across age groups. LSD
test was adopted to examine statistical significances between
each two of the groups and no significance was found.
No consistent trend of ATPK as a function of driver age
was found for the systems. Considering ADASs performances
in specific traffic situations, older drivers had lower SBZA
ATPK than younger drivers on inter-city highways. See
Figure 11. Besides, it seems that older drivers received
more FCWs on city expressways and more LDWs on city
roads and inter-city highways, which needs further verifi-
cations.
As indicated in Figure 12, FCW acceptance differed
among drivers due greatly to driver age (p=0.027). Younger
drivers favored FCW more than older drivers. Significant
difference were found to strengthen this finding (young and
old: p=0.038). In the evaluations of all the three systems,
young drivers’ acceptance rate ranked the highest among
the age groups. In general, SBZA got the highest evaluation
score among the three systems in any age group.
3.4. Driver Aggression
According to their aggression scores, subjects were divided
into three groups: prudent (0 to 55), moderate (56 to 70),
and aggressive (71 to 90). No one got a score higher than
90. The mean aggression score for each group was 33.9
(SD= 16.1), 61.5 (SD=3.3), and 79.1 (SD = 6.5), respec-
tively.
As expected, male drivers (Mean= 65.9, SD=18.5) were
more aggressive than female drivers (Mean= 48.9, SD=
21.4). Statistical significance was found between genders
(F(1,28)=5.231, p=0.030). This finding was consistent
with the results found by Reason et al. (1990). Considering
driver age, people tended to be more prudent as their age
increased (F(2,27)=1.269, p=0.297), consistent with the
results found by Lajunen and Parker (2001). The mean
aggression scores for young, middle, and old groups were
64.1 (SD=16.2), 63.2 (SD= 19.2), and 50.4 (SD=26.2),
respectively.
The relationships between ADASs performance and
driver aggression index varied across systems. The overall
performance of FCW showed that alert times increased
with driver aggression level (p=0.120). Unlike FCW, both
SBZA and LDW alert times had poorer correlation with
driver aggression levels. See Figure 13.
When comparing FCW ATPK across aggression groups,
Figure 10. Alert times per driver as a function of driver age.
Figure 11. ATPK in different traffic situations as a function
of driver age.
Figure 12. Driver acceptance as a function of driver age.
746 G. LI, S. EBEN LI and B. CHENG
aggressive drivers triggered FCW more frequently than the
other two groups. See Figure 14. The mean FCW ATPK on
city expressways were 1.6 (SD = 3.1), 1.0 (SD = 1.3), and
4.3(SD=1.3) for prudent, moderate, and aggressive drivers
respectively. The values were 2.2 (SD=1.4), 4.0 (SD=3.4),
and 5.3 (SD=5.9) on inter-city highways. Statistical signi-
ficance was found on city expressways (p=0.003). Multiple
comparison results presented more details (prudent and
moderate: p= 1.000; prudent and aggressive: p=0.021;
moderate and aggressive: p=0.007). No significance was
found for either SBZA or LDW.
As indicated in Figure 15, prudent drivers’ acceptance
rate was the highest among the aggression groups in the
evaluation of any ADAS. Differences between moderate
and aggressive groups did not consist across the subjective
rating on ADASs. No statistical significance was found.
4. DISCUSSIONS
4.1. Driver Acceptance and Driver Age
Consistent with the result found in this study, Ervin et al.
found that FCW acceptance differed among drivers due
largely to age, but contrarily, what Ervin et al. (2005) found
was older drivers viewed the FCW more favorably than
either middle or young drivers. As noted in this study,
drivers can change sensitive setting of the FCW and it was
found that older drivers preferred the most-sensitive setting
significantly more frequently than the other two groups
did. However, the evaluated FCW in this study shared the
same algorithm across all subjects. When the FCW could
not meet older drivers’ expectations, they may fully get
back to rely on their accumulated driving experience and
take the additional alerts as nuisances. This may be the
reason leading to the contrary results in these two studies.
4.2. Driver Acceptance and Driver Aggression
As indicated in Figure 15, prudent drivers’ acceptance rate
was always the highest among the aggression groups, while
the ranking of moderate and aggressive group did not keep
consistent. That’s probably because of the aggression group
categorization. Driver aggression score could range from 0
to 144, but gathered from 61 to 89. The mean aggression
scores for prudent, moderate, and aggressive groups was
33.9 (SD= 16.1), 61.5 (SD=3.3), and 79.1 (SD =6.5), respec-
tively. As indicated by the data distribution, moderate
drivers may have similar subjective feedback as Aggressive
drivers do. This probably may lead to the non-significant
results found between the aggression groups in this study.
Drawn from observations on typical aggressive drivers in
this study, they do like to change lanes frequently to over-
take other vehicles, drive over speed limits, and follow a
lead vehicle closely. All these behaviors are likely to trigger
an ADAS alert. For further analysis, more aggressive drivers
Figure 13. ADASs performance as a function of driver
aggression.
Figure 14. ATPK in different traffic situations as a function
of driver aggression.
Figure 15. Driver acceptance as a function of driver aggression.
FIELD OPERATIONAL TEST OF ADVANCED DRIVER ASSISTANCE SYSTEMS IN TYPICAL CHINESE 747
are needed to clarify the results.
4.3. ATPK
As indicated in Figure 5 and Table 4, higher speed traffic
situations would lead to significantly more FCWs and
LDWs, but fewer SBZAs. Statistical significances found in
driver gender, age, and aggression groups strengthened this
finding. See Tables 5, Table 6, and Table 7. Similar situ-
ations happened in driver acceptance results. Consistently,
SBZA ranks the first and LDW ranks the last in any group
of driver gender, age, or aggression. See Figures 9, Figure
12, and Figure 15. This complies with the results shown in
Figure 6.
4.4. False Alerts
As can be deduced from the results, a driver would receive
at least 3.3 false alerts per 100 km regardless of LDW.
However, in a study conducted by Sayer et al. (2011), the
average rate of false alerts for all warning types across
subjects was just about 0.5 per 100 km, 15% of the number
in this study. The relatively higher alarm rate of ADASs in
this study may be caused by the complex traffic situations
and typical road infrastructures in China. The rapidly
expanded Chinese automotive market and the lack of roads
and road infrastructures tremendously lead to serious traffic
situations in China. This greatly challenges the effective-
ness and acceptance of ADASs which may never happen in
US or Europe. Concerning road infrastructures, metal
guardrails exist almost everywhere by the sides of a road in
China, far more than that abroad. The guardrails would
easily trigger a SBZA when a driver is driving in the left or
right most lane. This aspect of system performance would,
in a way, negatively influence driver acceptance of SBZA.
4.5. Driver Feedback
Although the assessment of driver acceptance comes from
questionnaire responses overwhelmingly, extra communi-
cations on the advantages and disadvantages of ADASs
have added a degree of clarity in seeking to explain the
results lying behind the performance and driver acceptance.
From the perspective of ADASs advantages, they help
drivers to avoid crashes. Drivers can never detect every-
where in the blind zone. If they want to know more about
the traffic situations in the blind zone, they have to look
over their shoulders every time they want to make a lane
change. This would take them more attention and make
them easily get tired. In China, as there is no legal require-
ments to look over shoulders when changing lanes, most
Table 5. Significance test results and multiple comparisons results for gender groups.
Mean Significance
test result Pairwise comparisions results (p value)
CR CE ICH pCR & CE CR & ICH CE & ICH
FCW
ATP K
Male 0.0 4.2 6.2 <0.001 <0.001 <0.001 0.685
Female 0.0 0.4 3.6 <0.001 0.378 <0.001 0.001
SBZA
ATP K
Male 42.8 29.1 6.4 <0.001 0.428 <0.001 <0.001
Female 48.6 24.2 4.1 0.001 0.194 <0.001 0.011
LDW
ATP K
Male 15.8 30.3 36.8 0.001 0.009 <0.001 0.230
Female 12.2 28.4 43.0 0.019 0.060 0.006 0.377
Table 6. Significance test results and multiple comparisons results for age groups.
Mean Significance test
result Pairwise comparisions results (p value)
CR CE ICH pCR & CE CR & ICH CE & ICH
FCW
ATP K
Young 0.0 1.6 5.6 0.004 0.032 0.001 0.248
Middle 0.0 3.4 5.7 <0.001 0.010 <0.001 0.162
Old 0.0 4.3 4.4 0.003 0.006 0.001 0.687
SBZA
ATP K
Young 45.9 31.0 6.8 0.005 0.476 0.002 0.016
Middle 42.1 20.7 5.6 0.002 0.348 <0.001 0.012
Old 47.8 32.7 4.2 0.001 0.332 <0.001 0.009
LDW
ATP K
Young 10.3 29.0 34.5 0.078 0.090 0.031 0.594
Middle 14.0 27.0 38.1 0.004 0.057 0.001 0.111
Old 20.3 42.5 45.0 0.148 0.110 0.077 0.852
748 G. LI, S. EBEN LI and B. CHENG
Chinese drivers just look at the side mirrors to decide if it is
safe to do that. But the ‘did not see’ errors may lead to a
crash. The SBZA assist them in such situations and decrease
their workload. This leads to drivers’ preference for SBZA
assistance. According to traffic regulations in China, the
driver in the following vehicle would have to take all the
penalties and loss when a rear-end crash happened. FCW
can remind a driver when he/she follows a lead vehicle too
closely. This may prevent drivers from being involved in a
rear-end crash. Driver drowsiness can lead to a lane
departure while driving. When a vehicle departs from a
lane, the LDW can help alert the driver to avoid a potential
crash. Even if the driver is quite awake, a proper LDW
would help to keep in the center of the lane to avoid
conflicts with vehicles in adjacent lanes.
Feedback on ADASs disadvantages could help find
problems. Among the complaints on FCW, alert nuisances
caused by trucks or buses in adjacent lanes on inter-city
highways rank the first. For SBZA, false alters caused by
guardrails should be solved properly. Besides, passing-by
vehicles caused alerts may also be viewed as nuisance
when a driver was going straight without lane change
intention. Concerning the highest ranking complaints on
LDW, alerts were not expected to be triggered when a
driver was driving straight, far from riding the lane markers.
When ADASs alert drivers to more actual threats, their
opinions of ADASs will be more positive. However, if
drivers do not experience many actual treats, negative
opinions will accumulate, resulting from false alerts that
are deemed excessive or recurring (Najm et al., 2006). To
improve ADASs performance and driver acceptance in
China, the nuisances caused by trucks/buses, road infra-
structures, and other environment factors have to be solved.
Additionally, driver lane change maneuver needs to be
recognized more precisely and timely, and driver reaction
time in various traffic situations has to been take into
account (Li et al., 2014).
5. CONCLUSION
From the perspective of overall ADASs performance and
driver acceptance, LDW was triggered far more frequently
than either FCW or SBZA in Chinese typical road condi-
tions, while getting the lowest acceptance rate among the
systems. Alert frequencies of FCW and LDW increased in
higher speed traffic situations, whereas that of SBZA
declined. Subjective rating results showed that Chinese
drivers’ most favorable system was SBZA, with FCW
ranking the second and SBZA being the last. Similar trends
were found in each of the gender, age, or aggression groups.
Considering gender effect, male drivers received more
FCW alerts but fewer LDW alerts than female drivers, and
female drivers rated each ADAS more positively than male
drivers. In terms of driver age, older drivers received fewer
SBZA alerts than younger drivers did. Among the age
groups, young drivers’ subjective ratings ranked the highest.
Besides gender and age, driver aggression also showed
capability to affect ADASs performances and driver accep-
tances. The more aggressive a driver was, the more FCW
alerts he/she would receive. The observed acceptance rate
of prudent drivers ranked the highest in the evaluation of
each ADAS. These findings should be helpful for the
development of future ADASs for automotive industry.
However, this study is limited to short-term exposure with
the ADASs. This may not yield enough comprehensive
information for drivers to get adapted with the systems. For
future studies, longer exposure will be conducted.
ACKNOWLEDGEMENT−This study is partially supported by
the NSF of China under Grant 51205228, and Tsinghua University
Initiative Scientific Research Program under Grant 2012THZ0.
REFERENCES
Adell, E., Várhelyi, A. and Fontana, M. D. (2011). The
Table 7. Significance test results and multiple comparisons results for aggression groups.
Mean Significance
test result Pairwise comparisions results (p value)
CR CE ICH pCR & CE CR & ICH CE & ICH
FCW
ATP K
Prudent 0.0 1.6 2.2 0.002 0.104 <0.001 0.063
Moderate 0.0 1.0 4.0 0.001 0.096 <0.001 0.050
Aggressive 0.0 5.0 5.2 0.001 0.001 0.001 0.990
SBZA
ATP K
Prudent 49.3 28.8 5.5 0.003 0.223 0.001 0.028
Moderate 56.1 37.0 6.9 0.004 0.712 0.002 0.007
Aggressive 33.2 24.3 6.1 0.001 0.537 <0.001 0.003
LDW
ATP K
Prudent 8.7 36.1 39.5 0.036 0.036 0.018 0.781
Moderate 20.4 24.4 39.7 0.065 0.642 0.027 0.082
Aggressive 13.4 31.3 38.2 0.009 0.027 0.003 0.382
FIELD OPERATIONAL TEST OF ADVANCED DRIVER ASSISTANCE SYSTEMS IN TYPICAL CHINESE 749
effects of a driver assistance system for safe speed and
safe distance–A real-life field study. Trans. Research
Part C: Emerging Technologies 19, 1, 145−155.
Alkim, T. P., Bootsma, G. and Hoogendoorn, S. P. (2007).
Field operational test “The assisted driver”. Proc. 2007
IEEE Intelligent Vehicles Symp., IEEE, Istanbul, Turkey,
1198−1203.
Brookhuis, K. A., De Waard, D. and Janssen, W. H. (2001).
Behavioural impacts of advanced driver assistance systems–
An overview. European J. Transport and Infrastructure
Research 1, 3, 245−253.
Buss, A. H. and Perry, M. (1992). The aggression question-
naire. J. Personality and Social Psychology 63, 3, 452−
459.
Cronbach, L. (1951). Coefficient alpha and the internal
structure of tests. Psychometrica, 16, 297–334.
Deffenbacher, J. L., Getting, E. R. and Lynch, R. S. (1994).
Development of a driving anger scale. Psychological
Reports 74, 1, 83−91.
Dragutinovic, N., Brookhuis, K. A., Hagenzieker, M. P.,
and Marchau, V. A. W. J. (2005). Behavioural effects of
advanced cruise control use–A meta-analytic approach.
European J. Transport and Infrastructure Research 5, 4,
267−280.
Ervin, R., Sayer, J., LeBlanc, D., Bogard, S., Mefford, M.,
Hagan, M. and Winkler, C. (2005). Automotive Collision
Avoidance System Field Operational Test Report: Metho-
dology and Results. Technical Report DOT HS 809 900,
Department of Transportation, National Highway Traffic
Safety Administration, Washington, DC, USA.
Floudas, N., Amditis, A., Keinath, A., Bengler, K. and Engeln,
A. (2004). Review and Taxonomy of IVIS/ADASs Ap-
plications. Deliverable D2.1.2 of the EU-Project AIDE
Adaptive Integrated Driver-vehicle Interface. In the Frame
of the IST (Information Society Technologies) Program.
Golias, J., Yannis, G. and Antoniou, C. (2002). Classification
of driver-assistance systems according to their impact on
road safety and traffic efficiency. Transport Reviews 22,
2, 179–196.
Jagtman, H. M., Marchau, V. A. W. J. and Heijer, T. (2001).
Current knowledge on safety impacts of Collision
Avoidance Systems (CAS). Proc. Critical Infrastructures -
the 5th Int. Conf. Technology, Policy and Innovation,
Delft, The Netherlands.
Jeong, H. and Green, P. (2012). Forward Collision Warning
Modality and Content: A Summary of Human Factors
Studies. Technical Report UMTRI-2012-35, University
of Michigan Transportation Research Institute, Ann
Arbor, Michigan, USA.
Joy, M. (2007). Research Methods in Education. 10th edn.
The Higher Education Academy. York. UK.
Lajunen, T. and Parker, D. (2001). Are aggressive people
aggressive drivers? A study of the relationship between
self-reported general aggressiveness, driver anger and
aggressive driving. Accident Analysis and Prevention
33, 2, 243−255.
LeBlanc, D., Sayer, J., Winkler, C., Ervin, R., Bogard, S.,
Devonshire, J., Mefford, M., Hagan, M., Bareket, Z.,
Goodsell, R. and Gordon, T. (2006). Road Departure
Crash Warning System Field Operational Test: Metho-
dology and Results. Technical Report UMTRI-2006-9-1,
University of Michigan Transportation Research Institute,
Ann Arbor, Michigan, USA.
Lee, K. and Peng, H. (2005). Evaluation of automotive forward
collision warning and collision avoidance algorithms.
Vehicle System Dynamics 43, 10, 735−751.
Lee, M. S. and Jeong, H. Y. (2014). Driver propensity
characterization for different forward collision warning
times. Int. J. Automotive Technology 15, 6, 927–936.
Li, G., Wang, W., Li, S. E., Cheng, B. and Green, P. (2014).
Effectiveness of flashing brake and hazard systems in
avoiding rear-end crashes. Advances in Mechanical Eng-
ineering, Article ID 792670, DOI:10.1155/2014/792670.
Li, S., Li, K., Wang, J., Lian, X., Ukawa, H. and Bai, D.
(2010). Modeling of heavy-duty truck driver’s car-follow-
ing characteristics and its comparative verification. Int.
J. Automotive Technology 11, 1, 81−87.
Lindgren, A. and Chen, F. (2006). State of the art analysis:
An overview of advanced driver assistance systems
(ADASs) and possible human factors issues. Human
Factors and Economics Aspects on Safety, 38−50.
Ministry of Public Security Traffic Management Bureau
(2011). White Paper on Road Traffic Accidents in China
2010. P.R.C. Ministry of Public Security, Traffic Mana-
gement Bureau, Beijing, China. (Chinese).
Najm, W. G., Stearns, M. D., Howarth, H., Koopmann, J.
and Hitz, J. (2006). Evaluation of an Automotive Rear-
end Collision Avoidance System. Technical Report DOT
HS 810 569, Department of Transportation, National
Highway Traffic Safety Administration, Washington, DC,
USA.
National Transportation Safety Board (2001). Special Investi-
gation Report-Highway Vehicle and Infrastructure-based
Technology for the Prevention of Rear-end Collisions.
NTSB Number SIR-01/01, NTSB, Washington, DC, USA.
NHTSA (2011). NHTSA Vehicle Safety and Fuel Economy
Rulemaking and Research Priority Plan 2011−2013.
NHTSA–2009–0108, Department of Transportation,
National Highway Traffic Safety Administration, Wash-
ington, DC, USA.
NHTSA (2012). Traffic Safety Facts 2010: A Compilation
of Motor Vehicle Crash Data from the Fatality Analysis
Reporting System and the General Estimates System.
Technical Report DOT HS 811 659, Department of
Transportation, National Highway Traffic Safety Admini-
stration, Washington, DC, USA.
Oikawa, S., Matsui, Y. and Sakurai, T. (2014). Characteri-
stics of collision damage mitigation braking system for
pedestrian protection. Int. J. Automotive Technology 15,
7, 1129–1136.
Piao, J. and McDonald, M. (2008). Advanced driver as-
sistance systems from autonomous to cooperative ap-
750 G. LI, S. EBEN LI and B. CHENG
proach. Transport Reviews: A Transnational Transdiscip-
linary J. 28, 5, 659−684.
Reason, J., Manstead, A., Stradling, S., Baxter, J. and Campbell,
K. (1990). Errors and violations: A real distinction?.
Ergonomics 33, 10-11, 1315–1332.
Saad, F. (2004). Behavioural adaptations to new driver
support systems: Some critical issues. Proc. IEEE Int.
Conf. Systems, Man and Cybernetics. IEEE, The Hague,
The Netherlands, 288−293.
Sayer, J. R., Bogard, S. E., Buonarosa, M. L., LeBlanc, D.
J., Funkhouser, D. S., Bao, S., Blankespoor, A. D. and
Winkler, C. B. (2011). Integrated Vehicle-based Safety
Systems Light-vehicle Field Operational Test Key Find-
ings Report. Technical Report UMTRI-2010-21, University
of Michigan Transportation Research Institute, Ann Arbor,
Michigan, USA.
Shaout, A., Colella, D. and Awad, S. (2011). Advanced
driver assistance systems-past, present and future. Proc.
7th IEEE Int. Conf. Computer Engineering, IEEE, Giza,
Egypt, 72−82.
Suzuki, K. and Jansson, H. (2003). An analysis of driver’s
steering behaviour during auditory or haptic warnings
for the designing of lane departure warning system.
JSAE Review 24, 1, 65−70.
Xie, C. Q. and Parker, D. (2002). A social psychological
approach to driving violations in two Chinese cities.
Transportation Research Part F: Traffic Psychology and
Behaviour 5, 4, 293−308.
Zador, P. L., Krawchuk, S. A. and Voas, R. B. (2000). Auto-
motive collision avoidance system (ACAS) program.
Technical Report DOT HS 809 080, Department of
Transportation, National Highway Traffic Safety Admini-
stration, Washington, DC, USA.
Zhang, W., Huang, Y. H., Roetting, M., Wang, Y. and Wei,
H. (2006). Driver's views and behaviors about safety in
China-What do they NOT know about driving?. Accident
Analysis and Prevention 38, 1, 22−27.
