You are on page 1of 9

IATSS Research 42 (2018) 221–229

Contents lists available at ScienceDirect

IATSS Research

Research article

The public's understanding of the functionality and limitations of


automatic braking in Japan
Kan Shimazaki a,⁎, Tasuku Ito b, Ai Fujii c, Toshio Ishida d
a
Waseda University, Faculty of Human Sciences, 2-579-15, Mikajima, Tokorozawa City, Japan
b
Saitama Prefecture, Japan
c
Graduate School of Human Sciences, Waseda University, 359-1192 Tokorozawa City, Japan
d
Nihon University, College of Industrial Technology, 275-8575 Narashino City, Japan

a r t i c l e i n f o a b s t r a c t

Article history: Automated driving systems are increasingly widespread in the automotive industry. To operate automobiles that
Received 17 February 2017 are assisted by such systems, drivers must correctly understand their functionality and limitations. Hence, focus-
Received in revised form 31 October 2017 ing on the rapidly spreading technology of automated braking, this study seeks to clarify the general public's un-
Accepted 9 November 2017
derstanding of automatic braking systems. A total of 210 Japanese people responded to a questionnaire regarding
Available online 22 November 2017
how automatic brakes are operated, how they work, when they disengage, and so on. The proportion of those
Keywords:
who misunderstood the operation of the system is not especially high, but there were a certain percentage of
Automatic braking people who had misconceptions; e.g., some believed that automatic braking could detect pedestrians or bicycles.
Public understanding Moreover, respondents were under the impression that drivers cannot readily disengage automatic braking sys-
Functionality limitations tems. If they mistakenly use the automatic brake, there is a possibility of posing a risk to the road transportation
Questionnaire system.
© 2017 International Association of Traffic and Safety Sciences. Production and hosting by Elsevier Ltd. This is an
open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction The crash of China Airlines Flight 140 at Nagoya Airport was con-
cluded to have been caused by operating conflicts between the piloting
In recent years, a variety of automated driving systems have been staff and the onboard automated systems [2]. This demonstrates that
implemented, and they continue to be installed in many commercial ve- even well-trained, professional pilots can fail to completely understand
hicles. The USA's National Highway Traffic Safety Administration clas- the functionality and limitations of automated systems. The automated
sifies the degree of vehicle automation into six levels (ranging from 0 systems installed in automobiles are not as complex as those in aircraft;
to 5) based on the Society of Automotive Engineers (SAE) international however, drivers receive fewer hours of training than pilots do, and the
definitions [1]. Automated systems can handle all driving functions vast majority of drivers are non-professionals. Therefore, it would be
under certain conditions at Level 4 and under all conditions at Level 5, improper to presume that all drivers have a correct understanding of
but at Levels 1–3 (the current stage of practical application), they can the functionality and limitations of automated systems. This back-
only assist the driver, who remains in primary control of the vehicle. ground suggests that designers of automated systems need to know
When driving with assistance from such systems, it is essential for how well the vast majority of current drivers understand the function-
drivers to correctly understand their limitations and how errors can ality and limitations of such automated systems.
arise in their operation. If not, drivers may fail to predict the behavior To date, several polls on automated systems have been conducted.
of these systems, or assume they operate under conditions in which Bekiaris et al. [3] surveyed user needs and their acceptance of technolo-
they do not. The possibility of neither the driver nor the system taking gy that could assist drivers who were impaired. A questionnaire was
action is especially dangerous because the former assumes the latter conducted on 407 people in nine European countries. The survey re-
has a certain feature that is not present. vealed that most users would welcome being warned by a supportive
assistance system, but would not accept automatic driving. Howard &
Dai [4] investigated public attitudes toward self-driving cars using the
responses of 107 likely adopters in Berkeley, California, as a case
⁎ Corresponding author at: Faculty of Human Sciences, Waseda University, 2-579-15
study. They discovered that individuals are most attracted to potential
Mikajima, Tokorozawa City, Saitama, Japan.
E-mail addresses: shimazakikan@gmail.com (K. Shimazaki), mi6theuk@toki.waseda.jp safety benefits, the convenience of not having to find parking, and ame-
(T. Ito), fujii.ai@nihon-u.ac.jp (A. Fujii), ishi@waseda.jp (T. Ishida). nities such as multitasking while en route. Payre et al. [5] analyzed a
Peer review under responsibility of International Association of Traffic and Safety Sciences. priori acceptability, attitudes, personality traits, and intention to use a

https://doi.org/10.1016/j.iatssr.2017.11.002
0386-1112/© 2017 International Association of Traffic and Safety Sciences. Production and hosting by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
222 K. Shimazaki et al. / IATSS Research 42 (2018) 221–229

fully automated vehicle among 421 French drivers. It was found that Moreover, while predicting the movement of others, not only drivers
68.1% accepted fully automated driving, and the predictors of intention but also pedestrians and cyclists who do not have a license but use the
to use a fully automated car were mainly attitudes, contextual accept- road transportation system should be taken into account. Pedestrians
ability, and interest in impaired driving, followed by driving-related and cyclists will not be able to predict the movement of cars without
sensation seeking (all variables had a positive β) and gender (male an accurate understanding of automatic brakes. Therefore, we believe
had stronger intention). Schoettle & Sivak [6,7] examined public opin- that the survey on the functional understanding of automatic braking
ion regarding self-driving-vehicle technology among 1533 individuals should cover all road-using citizens. As we seek to reveal the percep-
in the US, UK, and Australia. Most respondents were positive regarding tions of these systems held by society as a whole, we enrolled partici-
the technology, and had high expectations about its benefits. However, pants regardless of whether they held a driver's license. As a result,
they expressed significant concern regarding security issues and self- approximately 15% of survey respondents did not hold a driver's license.
driving vehicles not performing as competently as actual drivers. Given this background, the research questions were as follows: (1)
Cyganski et al. [8] examined user acceptance and the perceived advan- What proportion of people believe that automatic braking has features
tages of automated vehicles on the basis of four specific cases within that do not actually exist? (2) What proportion of people believe that
the course of a quantitative online survey in Germany. The survey re- automatic braking does not have features that do actually exist?
vealed that perceived benefits were primarily window gazing and
relaxing, whereas the possibility of working seemed to be less valued. 2. Research design
Kyriakidis et al. [9] conducted surveys on 5000 people living in different
countries, and examined how variables such as residential area, age, 2.1. Automated systems of interest
gender, and personality influenced their acceptance of automated cars,
concerns, and purchase motivation. These studies clarify the public's ac- This study focuses on the automatic emergency braking (AEB) sys-
ceptance, concern, willingness, and popularization forecast regarding tem. An explanation of the AEB system can be found on the NHTSA
automation systems. Little research, however, has investigated how sat- and EURO NCA websites. Using radar and cameras, an AEB system de-
isfactorily the public comprehend the functionality and limitations of tects obstacles ahead of the host vehicle (and, in some cases, behind
automated systems. and to the side of it), and automatically applies the vehicle's brakes if
The aim of this research is not to clarify the acceptance, concern, necessary. Some systems brake until the vehicle comes to a complete
willingness, and popularization forecast of the public regarding automa- stop; others only slow down to reduce the impact of collision, without
tion systems. Rather, our aim is to clarify what people consider automat- coming to a halt. AEB systems vary from manufacturer to manufacturer
ed systems are able to do (or not do). In other words, we attempt to in terms of not only their product names, but also how they detect ob-
elucidate the probability of automatic braking misuse. Wahlström stacles, what obstacles they can detect, and the operating speeds at
[10], who investigated whether Finnish citizens would accept the auto- which they function (see appendix). People may be able to name specif-
mation of Helsinki Metro, is one of the few researchers to question the ic automotive manufacturers and product names, but very few have an
public understanding of an automated system. In this research, question accurate understanding of the details of the different automatic braking
items enquired as to whether “Sensors may cause problems (vandal- systems. Accordingly, we decided to investigate general perceptions of
ism/unnecessary stops)” and “It is unknown how the sensors will AEB as a technology, rather than those of specific manufacturers or
work in harsh Finnish weather conditions.” These question items ex- products. The Japanese Ministry of Land, Infrastructure and Transport
plore how well people understand the behavior of the automated sys- uses the official name “Collision damage mitigation braking” for AEB.
tem. However, opinions on automation between railway passengers However, ordinary people refer to AEB as “automatic braking,” and
and car drivers should be fundamentally different. Passengers do not di- this name is also used in advertisements. Therefore, in this study, we
rectly interact with the Metro's automated system. Conversely, the driv- use the term “automatic braking” for AEB. Our survey asked participants
er must cooperate with the automation system, at least below Level 3 of about their perceptions of this technology using the term “automatic
the SAE. A survey conducted by Continental AG in Germany, China, braking systems,” without explaining the above methodological
Japan, and the US [11] indicated that several people doubted the reli- consideration.
ability of the automated driving function. However, it only questioned
“reliability” in general, not which specific automation system functions 2.2. Survey participants & survey period
were reliable.
To address this gap in the literature, the present study focuses on au- To develop a braking system that ensures adequate braking force re-
tomatic braking, a technology that has spread rapidly in recent years gardless of how hard the driver steps on the brake pedal, it is necessary
and has received significant media coverage. Automatic braking sys- to clarify the distribution of the treading force of several people and de-
tems are considered to contribute greatly to safety, more so than other sign it to suit weak treading forces. The purpose of this research follows
automated systems like lane-keeping and preceding-vehicle-following a similar logical structure. To develop an automatic braking system that
technologies. Recent years have seen these systems continuing to is safe for anyone to use, it may be necessary to clarify the distribution of
spread from cars to large vehicles like trucks and buses; in Japan, understanding of many people and develop the system according to the
there is a movement to mandate that all new cars be equipped with person with the weakest understanding. At the same time, such data
the technology. Thus, we can expect automatic braking systems to be- also provide an estimate of how much education drivers require. For
come more prevalent and improve over time. In this study, we attempt this reason, our research targets many people. Therefore, detailed inter-
to clarify the public's understanding of automatic braking systems. By views of small groups of people were not conducted.
doing so, it will be possible to provide automobile developers with valu- Our survey was conducted in Japan through Macromill Inc., a com-
able information such as predictions of misuse. In addition, this knowl- pany that provides internet survey services. Macromill Inc. is one of
edge can direct the extent to which car dealers should explain the the largest internet research companies in Japan, with 2.3 million survey
technology to purchasers. Furthermore, it guides policy makers on monitors registered. Survey participants were 20 years of age and older.
how much technical information should be provided to the public. People working in the automotive industry were excluded, including
There is a notion that only the drivers who use cars would under- automobile manufacturers, dealers, and automotive repair servicers, be-
stand the function of automatic braking. Based on this notion, it might cause these people may be more familiar with the function of automatic
be thought that only licensed drivers should participate in this survey. braking systems than the general public. In total, 210 participants were
However, traffic behavior in the road transport system is interactive. selected at random, consisting of 21 men and 21 women in each of the
This includes the process of predicting the movement of other drivers. following age groups: 20–29, 30–39, 40–49, 50–59, and ≥60 years old.
K. Shimazaki et al. / IATSS Research 42 (2018) 221–229 223

The number of subjects to be surveyed was determined by the research The question items were determined by discussion between the au-
budget. Participants were enrolled without pre-conditions on area of thor and co-authors. Before starting this investigation, we asked several
residence, income, family structure, or driver's license ownership status. people who were unaware of the research purpose to answer the ques-
In other words, survey subjects were chosen at random, excluding peo- tions in order to ascertain whether the questions items could be
ple from the automobile industry. Therefore, the subjects of this survey understood.
can be assumed to be representative of the Japanese population, except
that people without access to the Internet could not participate. In total, 3. Results
84.3% (177/210) of participants held a driver's license, approximately
equivalent to the rate of driver's license ownership in Japanese adults Table 2 presents data on the respondents' experiences driving and
aged 20–74 (85.8%) [12]. Participants were divided into four driver riding in vehicles with automatic braking systems, and their exposure
groups based on their responses to questions about their possession of to information concerning such systems. Few respondents had an auto-
a license and driving frequency: “Non-license holder,” “Infrequent driv- matic braking system in their own car (6.2%), whereas 10.0% had seen a
er” (drivers who rarely drive at all), “Weekend driver” (drivers who demonstration or test-driven or ridden in such a vehicle. In total, 15.8%
drive 1–2 days a week), and “Heavy user” (drivers who drive ≥3 days of respondents answered “Yes” to at least one of these two items. A rel-
a week). Table 1 summarizes the number of individuals and gender atively high percentage of respondents had encountered information
breakdown of each driver group. The survey period was February about automatic braking systems. The medium responsible for the sin-
2015. Survey data received from Macromill Inc. were anonymized to gle highest exposure rate was television commercials, followed in de-
make it impossible for researchers to contact the participants. This scending order by the news, the Internet, newspaper advertisements,
study was granted a waiver by the Waseda University Ethics Committee and friends/acquaintances. In total, 86.3% of respondents answered
and did not require ethical approval. that they had heard about automatic braking through at least one
medium.
2.3. Questionnaire contents We analyzed the 23 six-choice question items according to the fol-
lowing procedure. First, we created histograms using individual re-
First, participants were asked whether their car was equipped with sponses to the six-choice items (Strongly disbelieve through Strongly
an automatic braking system, and whether they had seen a demonstra- believe; Fig. 1). Next, we calculated the average response value and
tion of, or taken a test drive or ride in, a car equipped therewith. Partic- standard deviation for each question item. We assigned point values
ipants were also asked whether they had encountered information to responses as follows: Strongly disbelieve = −2.5, Moderately disbe-
related to automatic braking systems before, with separate questions lieve = −1.5, Slightly disbelieve = −0.5, Slightly believe = 0.5, Mod-
for different media sources of information (TV commercials, newspaper erately believe = 1.5, Strongly believe = 2.5. This scale set the
advertisements, the Internet, the news, and friends/acquaintances; see median response value at zero, midway between Slightly disbelieve
Table 2). These questions were asked to clarify the extent to which sur- and Slightly believe. Accordingly, the positive and negative average re-
vey subjects were exposed to information about automatic brakes. sponse scores indicate that respondents generally believed and
Automatic braking is a substitute for part of conventional driving be- disbelieved, respectively, the associated statement (Table 3, Column 2).
havior. Therefore, users of the automatic braking system will think that We next conducted a two-way ANOVA for each question item to
automatic braking works in a manner similar to the driver's behavioral check for differences in responses due to personal characteristics. The
model. Moreover, a vehicle equipped with an automated system is in a response value was used as the dependent variable, and the gender
state where two operators, a driver and an automation system, are on and driver groups in Table 1 were used as the two factors. The main ef-
board. Therefore, it is necessary to clarify the extent to which the driver fects of the driver group on response values that reached statistical sig-
can trust the automated system. In addition, it is necessary to clarify nificance were further analyzed using multiple comparisons (Table 3,
which operator is considered to be given priority if there is a mismatch Columns 3–4). There were no significant interactions between driver
between the decision of the driver and the automated system. Based on group and gender. To check for age-related differences, we also con-
these points, the following nine question items asked about what auto- ducted correlation analysis to identify significant correlations between
matic braking systems detect (Questions on “hazard perception” of the age and response value (Table 3, Column 5).
driving behavior model), five questions asked about how they operate For the final three items (whether the presence of an automatic
(Questions on “operation” of the driving behavior model including in- braking system made participants feel relieved, whether it made them
formation presentation), three questions asked about conditions inattentive, and whether they believed all vehicles should be equipped
under which they are not engaged (Questions related to priority), and with such systems), we hypothesized that participants' responses
three questions asked about system malfunctions (Questions about would be influenced by their knowledge and perceptions of automatic
the “accuracy of judgment” of the driving behavior model). Finally, par- braking systems as measured by question items 1–20. To confirm, we
ticipants were asked whether the presence of an automatic braking sys- conducted three multiple regression analyses using a stepwise method,
tem made them feel relieved, whether it made them inattentive, and each using the response score for one of these three items as the depen-
whether they believed all vehicles should be equipped with such sys- dent variable and the response score for items 1–20 as explanatory var-
tems (Questions about user trust of the system). Participants responded iables (Table 4).
to all 23 items with one of six responses: Strongly disbelieve, Moderate- Table 3 shows survey results for individual questions. For the ques-
ly disbelieve, Slightly disbelieve, Slightly believe, Moderately believe, tion about what automatic braking systems detect, the majority of re-
and Strongly believe. spondents answered that automatic braking systems are engaged
when they detect preceding vehicles (68.6%), stopped cars ahead
(84.3%), and walls (76.7%). In contrast, few believed they could detect
Table 1 cars coming from the side of the vehicle (21.9%), smoke (13.8%), stop
Attributes and numbers of survey respondents.
signs (11.4%), or red signal lights (17.1%). More people believed that au-
Attribute Notation (for multiple comparisons) Men Women tomatic braking systems do not detect bicycles (37.1%) or pedestrians
Non-license holders N (No license) 15 18 (42.9%) than disbelieved, but these items received more affirmative re-
People who rarely drive I (Infrequent driver) 15 30 sponses than did the items for stop signs (13.8%) and red lights (11.4%).
People who drive 1–2 d/wk W (Weekend driver) 36 26 Regarding how automatic braking systems operate, overall, people
People who drive ≥3 d/wk H (Heavy user) 39 31 believed that automatic braking systems function by reducing the
Unit = people. speed of a vehicle (58.1%) and completely stopping a vehicle (63.8%);
224 K. Shimazaki et al. / IATSS Research 42 (2018) 221–229

Table 2
Experience driving and riding in vehicles with automatic braking systems and information exposure.

Question Item Percentage responding “Yes” Category Percentage responding “Yes” to at least one item

Own car is equipped with AB system 6.2% Driving/riding experience 15.8%


Saw demonstration/took test drive or ride 10.0%
TV commercial 78.1% Information exposure 86.3%
Newspaper advertisement 20.5%
Internet 24.3%
News 42.4%
Friend or acquaintance 10.0%

few believed that such systems warn the driver through sound (34.8%) Gender-related differences in response score were observed for Q13
or a display (40.0%). Few respondents believed that automatic braking and Q22. For both items, the female score was significantly higher than
systems would turn the steering wheel to avoid obstacles (16.7%). the male score. Further, men believed more strongly than women that
Regarding the conditions under which automatic braking systems such systems completely stop vehicles. Men also believed more strongly
are disengaged, the mean response scores for all three items were neg- than women that the presence of such a system would make them
ative, indicating the public perceives it as being difficult for drivers to inattentive.
disengage automatic braking systems. There were significant driver group effects on the response scores
In terms of false positive and false negative operations by automatic for Q12, Q13, Q14, and Q21. Compared with other groups, infrequent
braking systems, the mean response score was essentially zero for the drivers believed less strongly that automatic braking reduces the
question item about false positive detections and malfunctions (i.e., au- speed of a vehicle, and believed more strongly that it brings vehicles
tomatic braking systems engaging when nothing is wrong). to a complete stop. Compared with non-license holders and infrequent
The final question items asked respondents whether the presence of drivers, weekend drivers felt more strongly that automatic braking sys-
an automatic braking system made them feel safer, whether it made tems would not make them feel safer.
them inattentive, and whether they believed all vehicles should be There were significant correlations between response score and age
equipped with such systems. For the first two questions, the responses for Q13, Q18, Q19, and Q20. Younger respondents believed more strong-
had mean scores of essentially zero, but their distributions about the ly that automatic braking systems bring vehicles to a complete halt.
mean were asymmetric. Few people felt strongly relieved (5.7%) or inat- Older participants estimated the risk of both false positive and false neg-
tentive (9.0%) in vehicles equipped with automatic braking systems, but ative engagement of automatic braking system as being lower, and be-
many believed they would be slightly relieved or slightly inattentive lieved more strongly that the installation of such a system on a vehicle
(31.4%, 33.3%). In contrast, roughly similar numbers of respondents an- would completely prevent collisions. In short, older respondents tended
swered that they strongly disbelieved (10.5%, 11.9%) and slightly to place greater trust in automatic braking.
disbelieved (18.1%, 20.5%) these two assertions. A majority of respon- We next investigated whether the responses to Q1–Q20 (Questions
dents (62.4%) believed that all vehicles should be equipped with such about automatic braking knowledge) had any effect on the responses to
systems; however, the variation in the response score was large. Q21, Q22, and Q23. Table 4 presents the results of three multiple

Fig. 1. Histograms of individual responses.


K. Shimazaki et al. / IATSS Research 42 (2018) 221–229 225

Table 3
Statistical analysis results of individual responses.

Question item Mean (SD) ANOVA main effects Correlation with age

Gender Driver group (Significance)

Questions about “hazard perception” of driving behavior model (what automatic braking systems detect)
Q1 Automatic braking systems engage when a moving car ahead is 0.53(1.39) n.s. n.s. r = 0.10 (n.s.)
detected
Q2 Automatic braking systems engage when a stopped car ahead is 1.09(1.12) n.s. n.s. r = 0.05 (n.s.)
detected
Q3 Automatic braking systems engage when a car coming from the −1.02(1.34) n.s. n.s. r = −0.02 (n.s.)
side is detected
Q4 Automatic braking systems engage when a bicycle is detected −0.44(1.48) n.s. n.s. r = 0.09 (n.s.)
Q5 Automatic braking systems engage when a pedestrian is detected −0.37(1.53) n.s. n.s. r = 0.03 (n.s.)
Q6 Automatic braking systems engage when a wall is detected 0.81(1.23) n.s. n.s. r = −0.07 (n.s.)
Q7 Automatic braking systems engage when smoke is detected −1.49(1.22) n.s. n.s. r = 0.03 (n.s.)
Q8 Automatic braking systems engage when a stop sign is detected −1.60(1.21) n.s. n.s. r = 0.05 (n.s.)
Q9 Automatic braking systems engage when a red signal light is −1.46(1.37) n.s. n.s. r = 0.06 (n.s.)
detected

Questions about “operation” of driving behavior model including information presentation (how automatic braking systems operate)
Q10 Automatic braking systems alert the driver via an audio warning if −0.62(1.4) n.s. n.s. r = 0.03 (n.s.)
an obstacle is detected
Q11 Automatic braking systems alert the driver via a display if an −0.40(1.35) n.s. n.s. r = 0.01 (n.s.)
obstacle is detected
Q12 Automatic braking systems apply the brakes to reduce speed if an 0.16(1.4) n.s. F = 3.80, p = 0.011⁎ (I b W, I b H) r = −0.12 (n.s.)
obstacle is detected
Q13 Automatic braking systems apply the brakes to completely stop 0.33(1.5) F = 5.41, p = 0.021⁎ F = 4.65, p = 0.004⁎⁎ (I N H) r = −0.13 (t = 1.98, p
the vehicle if an obstacle is detected (men b women) = 0.049⁎)
Q14 Automatic braking systems turn the steering wheel to avoid −1.41(1.28) n.s. F = 3.19, p = 0.025⁎ (n.s. on r = 0.02 (n.s.)
collision if an obstacle is detected multiple comaparison)

Questions about given priority (conditions under which automatic braking systems are disengaged)
Q15 Automatic braking systems do not engage if the driver is −0.65(1.27) n.s. n.s. r = 0.11 (n.s.)
accelerating
Q16 Automatic braking systems do not engage if the driver is braking −0.40(1.33) n.s. n.s. r = −0.03 (n.s.)
Q17 Automatic braking systems do not engage if the driver is turning −0.78(1.27) n.s. n.s. r = 0.00 (n.s.)
the steering wheel

Questions about the “accuracy of judgment” of driving behavior model (malfunction-related beliefs about automatic braking systems)
Q18 Automatic braking systems can accidentally engage when there is 0.08(1.42) n.s. n.s. r = −0.18 (t = 2.58, p
nothing wrong = 0.010⁎)
Q19 Automatic braking systems can fail to engage at times when they 0.43(1.41) n.s. n.s. r = −0.22 (t = 3.52, p
should b 0.001⁎⁎⁎)
Q20 Presence of an automatic braking system completely prevents −1.43(1.34) n.s. n.s. r = 0.20 (t = 3.00, p
collisions = 0.003⁎⁎)

Questions about user trust for the system (automatic braking-related emotions, behaviors, and beliefs)
Q21 Presence of an automatic braking system makes me feel safer 0.03(1.38) n.s. F = 4.22, p = 0.006⁎⁎ (N N W, r = 0.01 (n.s.)
I N W)
Q22 Presence of an automatic braking system makes me inattentive −0.07(1.42) F = 6.14, p = 0.014⁎ n.s. r = −0.10 (n.s.)
(men b women)
Q23 All vehicles should be equipped with automatic braking systems 0.38(1.48) n.s. n.s. r = 0.09 (n.s.)

Responses scored from −2.5 (Strongly disbelieve) to 2.5 (Strongly believe) in 1-point intervals.
⁎ 0.01 b p b 0.05.
⁎⁎ 0.001 b p b 0.01.
⁎⁎⁎ p b 0.001.

regression analyses performed using the respective scores for Q21, Q22, turning the steering wheel (Q17),” with a stronger belief predicting
and Q23 as dependent variables and the scores of the other 20 items as a greater feeling of safety.
explanatory variables. The multiple regression analysis was applied The score for the “Presence of an automatic braking system makes
using a stepwise method; the table shows each variable entered, along me inattentive” was significantly affected by two items: stronger beliefs
with each regression's β value and test statistics. In addition, the Vari- that “Automatic braking systems can fail to engage at times when they
ance Inflation Factor (VIF) was calculated to confirm that there were should” and “Automatic braking systems do not engage if the driver is
no multiple collinearities in the three multiple regression analyses. As braking.” There was also a marginally significant effect from the score
a result, it was confirmed that all VIFs were b1.88 (general removal for “Automatic braking systems alert the driver via a display if an obsta-
criteria for multicollinearity is VIF ≥ 10). cle is detected,” with a stronger belief predicting greater inattention.
The score for the “Presence of an automatic braking system makes The score for “All vehicles should be equipped with an automatic
me feel safer” was significantly affected by three items: stronger be- braking system” was significantly affected by three items: stronger be-
liefs that “Automatic braking systems engage when stopped cars liefs that “Automatic braking systems engage when stopped cars
ahead are detected (Q2),” “Presence of an automatic braking system ahead are detected,” “Presence of an automatic braking system
completely prevents collisions (Q20),” and “Automatic braking completely prevents collisions,” and “Automatic braking systems alert
systems can accidentally engage when there is nothing wrong the driver via an audio warning if an obstacle is detected.” There were
(Q18).” There was also a marginally significant effect from the also marginally significant effects from the scores for “Automatic brak-
score for “Automatic braking systems do not engage if the driver is ing systems apply the brakes to completely stop the vehicle if an
226 K. Shimazaki et al. / IATSS Research 42 (2018) 221–229

Table 4
Multiple regression analysis with emotions, behaviors, and opinions as dependent variables.

Dependent variable Introduced variable Statistics

Question item β Value

Q21 Presence of an automatic braking system makes me Q2 Automatic braking systems engage when a stopped car ahead is detected 0.52*** Rsq = 0.29
feel safer Q20 Presence of an automatic braking system completely prevents collisions 0.26*** Adj. Rsq =
Q18 Automatic braking systems can accidentally engage when there is nothing −0.17** 0.28
wrong F = 20.9
Q17 Automatic braking systems do not engage if the driver is turning the steering 0.12† p b 0.001
wheel
Q22 Presence of an automatic braking system makes me Q19 Automatic braking systems can fail to engage at times when they should 0.19** Rsq = 0.13
inattentive Q16 Automatic braking systems do not engage if the driver is braking 0.16* Adj. Rsq =
Q11 Automatic braking systems alert the driver via a display if an obstacle is detected 0.14† 0.11
Q20 Presence of an automatic braking system completely prevents collisions 0.12 n.s. F = 7.65
p b 0.001
Q23 All vehicles should be equipped with automatic Q2 Automatic braking systems engage when a stopped car ahead is detected 0.38*** Rsq = 0.19
braking systems Q20 Presence of an automatic braking system completely prevents collisions 0.19* Adj. Rsq =
Q10 Automatic braking systems alert the driver via an audio warning if an obstacle is 0.17* 0.16
detected F = 6.83
Q13 Automatic braking systems apply the brakes to completely stop the vehicle if an 0.16† p b 0.001
obstacle is detected
Q18 Automatic braking systems can accidentally engage when there is nothing −0.11†
wrong
Q14 Automatic braking systems turn the steering wheel to avoid collision if an −0.14
obstacle is detected n.s.
Q12 Automatic braking systems apply the brakes to reduce speed if an obstacle is −0.13
detected n.s.

p b .1
⁎ p b .05.
⁎⁎ p b .01.
⁎⁎⁎ p b .00.

obstacle is detected” and “Automatic braking systems can accidentally On the questions about disengaging the automatic brake, the majority
engage when there is nothing wrong,” with stronger belief and disbelief of respondents thought that the automatic braking system would not dis-
(respectively) predicting stronger belief in their necessity. engage even in the presence of driver operation. If automatic braking is con-
sidered as a technology for ensuring safety, it is natural to assume that such
4. Discussion a system would operate independently of driver input. In contrast, if auto-
matic braking is considered as a technology for promoting driver aware-
The majority of survey respondents knew of the existence of auto- ness, there is no need for it if the driver begins collision avoidance
matic braking systems, but their perceptions of this new technology maneuvers before the system. The fact that respondents considered auto-
were constrained by a lack of user experience. A large dispersion was matic braking systems difficult to disengage suggests that most people
observed in responses to the bicycle and pedestrian items, indicating hold the former belief. However, the driver is primarily in control of vehi-
large variation in public perceptions. Recent years have seen an increase cles graded as automation levels 0–2 in the SAE hierarchy; this means
in automatic braking systems that can detect pedestrians; however, this that automatic braking systems in commercial vehicles are designed pri-
fact does not yet appear to be well known. Fewer systems are intended marily under the latter assumption. For example, Toyota's automatic brak-
to detect bicycles than pedestrians, yet the belief that they do detect pe- ing system (Pre-Crash Safety™) specifies in its operating manual that the
destrians and the belief that they do detect bicycles had a similar prev- system does not engage when the driver is applying the brake or making
alence among respondents. There are currently no automatic braking other collision-avoidance maneuvers. Automatic braking systems should
systems designed to detect stop signs and red lights; indeed, few re- not be designed categorically based on one or other of these assumptions.
spondents believed that such systems detect these objects. While auto- However, automotive manufacturers would be better served by consider-
matic braking systems are not intentionally designed to detect smoke, it ing the two assumptions separately, and likewise user-oriented consider-
may sometimes be erroneously detected by systems that utilize cam- ations should not be mixed up with designer-oriented ones.
eras. (For example, the operating manual for Subaru's EyeSight™ Many more people believed there was a risk of false negative detec-
notes the possibility of accidental system engagement due to smoke, tions and operational failure (i.e., automatic braking systems failing to
and recommends the system be disabled in smoky places.) However, engage when they should). However, most participants did not consider
it appears that most participants did not consider this risk of “false pos- this risk of “false positive” operations by automatic braking systems.
itive” operations by automatic braking systems. These two items are in a trade-off relationship. In other words, raising
In reality, the automatic braking systems produced by most manu- the sensitivity of an automatic braking system to prevent missed detec-
facturers promote driver awareness by alerting them via audio or a dis- tions increases the rates of false positive detections and engagement,
play; however, this fact is apparently not yet widely known. In contrast, whereas lowering its sensitivity to prevent erroneous detections in-
63.8% people believed that automatic braking systems completely stop a creases the rates of false negative detections and engagement. Eliminat-
vehicle rather than merely slow it down, indicating the role of such sys- ing both possibilities is considered difficult, and so drivers must be able
tems is perceived more strongly as bringing vehicles to a halt rather to envision both situations. However, our results show that the public
than merely reducing their speed. If the conditions are right, some auto- finds it more difficult to imagine the erroneous triggering of automated
matic braking systems will indeed go so far as to stop the vehicle driving systems than their failure to engage. Few respondents thought
completely, but others will only slow to reduce damage from a collision. that the presence of an automatic braking system could completely pre-
The results suggest that the public may place too much trust in the func- vent collisions. This indicates that respondents trusted the technology a
tionality of such systems. little, but would not put complete faith in it.
K. Shimazaki et al. / IATSS Research 42 (2018) 221–229 227

Differences in gender and driver groups suggest that interest in auto- assumed that automatic braking systems possessed this feature. We
motive technologies is lower in people who drive infrequently, as well as can conceptualize the incredible risk of this situation by imagining a
in women (compared to men). This result about gender is consistent with world where even 1% of drivers on the road think that a red light
previous studies indicating that men are more likely to accept automated means “go.”
systems than women [5,9,13,14]. These groups may have stronger mental With technological progress, the capabilities and spread of automat-
images of automatic braking being for stopping rather than for reducing ed systems are improving every day. People's perceptions of these sys-
speed. This result about driver groups may be the cause of Missel's [15] tems will change as their capabilities improve, and their
findings that younger people and people who are “just not interested in understandings of their functionality and limitations will deepen with
cars” were more likely than other groups to view automated vehicles pos- increased opportunities for them to encounter automated systems in
itively [15]. real life. Therefore, we believe that researchers must continue to inves-
The results for the correlations with each item and age revealed that tigate the public's understanding of these technologies.
young people overestimated the function of automatic braking more than Overall, people having incorrect knowledge about automatic braking
older people. This can be explained by the fact that automatic braking sys- systems were never in the majority. However, the proportion of those
tems capable of stopping vehicles were introduced only recently, whereas who think “It is possible for automatic braking…” despite it actually being
the technology, in its infancy, could only reduce a vehicle's speed. We be- impossible, or the proportion of those who think “Automatic braking will
lieve that older people's perceptions of automatic braking were strongly not work with this kind of behavior” despite automatic braking actually
shaped during an era when the technology did not bring vehicles to a com- having such a function, are not sufficiently low. Automatic braking has dif-
plete stop, whereas younger people, born later and therefore largely igno- ferent functions depending on the manufacturer and era; hence, we cannot
rant of this era, are more cognizant of this functionality. This result seems strictly specify whether a person's knowledge is correct or not. However,
to be closely related to the telephone survey results reported by TE Connec- the incorrect use of automatic braking is considered a risk in the road trans-
tivity [16], which revealed that younger drivers feel more comfortable with portation system, even if these cases were relatively rare.
automatic driving than elderly drivers.
It is not hard to imagine why Q1 (AB system detects stopped car 5. Limitations
ahead), Q20 (AB system completely prevents collisions), and Q18 (AB
system can accidentally disengage; negative β) were significant in The survey was conducted over the Internet. Therefore, it cannot be
influencing the participants' feeling of safety (Q21). The responses to said that the respondents were completely representative of Japanese
Q17 (AB systems do not engage when turning the steering wheel) indi- citizens. However, in recent years, the percentage of the population
cate that people enjoy a greater feeling of safety when they believe that with ready access to the Internet has increased considerably, and this
driver intervention can readily disengage automatic braking, and this problem is unlikely to have significantly affected the results.
can be interpreted as a corollary to the anxiety drivers feel toward auto- Further, this survey only provided data on Japanese people. In Japan, the
mated systems having primary control over driving. population density is high, and the proportion of urban streets mixed with
The score for Q22 (AB system makes me inattentive) was significant- pedestrians and bicycles is higher than in many other countries. Further, as
ly affected by Q19 (AB systems can fail to engage at times when they Japan is a global producer of cars and is economically relatively wealthy, the
should) and Q16 (AB systems do not engage if the driver is braking). exposure to information on cars equipped with new technology is high.
There was also a marginally significant effect from the score for Q11 Therefore, the results of this research may be different in other countries,
(AB systems alert the driver via a display), with a stronger belief and similar surveys should be conducted in various countries.
predicting greater inattention. At first glance, these findings appear par- This problem applies not only to the differences in countries, but also
adoxical: people claim they are more inattentive both if they believe the to differences in eras and generations. With age, the performance of au-
automatic system will often fail to engage when it should and if they be- tomated systems will improve, the penetration rate will rise, and people
lieve primary driving control rests with the driver. The reason for this will have more opportunities to use them. Likewise, younger genera-
observation is unclear; it is possible that the relationship between tions have been exposed to digital information technology since birth,
these items is not a causal one. We posit that this inattentive tendency and their understanding may be deeper than that in other generations.
may be the result of some kind of common factor making people skep- Therefore, it is necessary to continuously conduct similar surveys in the
tical of automated systems, perhaps a result of both the respondents' future and to examine these changes.
personalities and their individual backgrounds (region, culture, etc.). As seen from the low R2 values, all three multiple regression analyses
Items introduced into the regression of Q23 (All vehicles should be had low explanatory power. We note that the three dependent variables
equipped with AB systems) were the same or similar in content to examined would be readily influenced by social desirability, possibly
those that affected the reported feeling of safety. We hypothesize that lowering the explanatory power of the regression. We must be careful
having high expectations of automatic braking leads drivers to believe to distinguish between someone self-reporting that the presence of an
that most vehicles should be equipped with the systems. automatic braking system makes them inattentive and real-world inat-
Readers should exercise caution when interpreting the results of the tentive behavior while driving. This topic needs to be further investigat-
present study. The mean response scores were calculated such that ed using actual driving data and other empirical methods.
values greater than zero indicate belief and values less than zero indi- Another viable explanation for these results can be posited: the re-
cate disbelief. However, readers must bear in mind that these are mere- sponses to these three items may only be slightly influenced by drivers'
ly average values. For example, none of the automatic braking systems awareness of the specific functionality of automated systems. Other fac-
currently on the market can detect red signal lights. The average score tors may have been responsible instead, e.g., respondents' individual ex-
for the associated questionnaire item was − 1.46, meaning that the periences (their own or those of friends), exposure to reports of reduced
vast majority of people believe that such systems do not detect red accident rates (due to such systems or other factors), and other ele-
lights. However, 2.9% of respondents strongly believed that automatic ments not examined in this study.
braking systems do detect red signal lights; when the respondents On the other hand, this result reaffirms the value of this research,
who moderately believed this statement are included, this value rises which attempted to directly clarify the knowledge of automation sys-
to 7.6%. In fact, 17.9% of respondents at least slightly believed it to be tems. This is because variables such as acceptance attitudes, concerns,
true. These were in the minority of respondents, and their misconcep- and trust in automated systems, as revealed by previous studies, are
tions would likely be rectified by the time they actually drove or rode only weakly linked with the knowledge of automated systems.
in a car equipped with an automatic braking system. However, it At the time of the survey, the penetration rate of automatic brakes
would be a serious situation if even a minority of drivers on the road was not especially high. Therefore, most of the participants of this
228 K. Shimazaki et al. / IATSS Research 42 (2018) 221–229

study did not own a car equipped with automatic brakes. The penetra- indicate that the vast majority of people have an essentially correct under-
tion rate of automatic braking is gradually rising. People who own cars standing, with few people reporting mistaken perceptions, such as auto-
with automatic brakes are likely to have better knowledge about their matic braking systems being able to completely prevent collisions. Few
operation. Even if someone owns a car equipped with automatic brakes, systems are intended to detect bicycles, yet the belief that they can was
incorrect knowledge could result in direct and serious danger. There- comparable in strength and spread to the belief that they can detect pedes-
fore, in the future, we would like to conduct a survey targeting people trians. Respondents were under the impression that drivers cannot readily
who have cars equipped with automatic braking. disengage automatic braking systems, and were less concerned with such
Drivers should have an accurate understanding of the automation systems falsely activating in inappropriate situations than with them failing
systems equipped on their cars. An accurate understanding not only re- to activate when they should. The proportion of those who misunderstood
duces the chance of misuse, but also prompts appropriate decisions in is not high, but there were a certain percentage of people who misunder-
circumstances where there is likely to be an accident. For example, in stood the functions and limitations of automatic braking. If they mistakenly
situations when driving through a light obstacle may prevent a collision use the automatic brake, there is a possibility of posing a risk to the road
with another vehicle, it is generally better to drive through the obstacle. transportation system. Respondents' beliefs about the functionality of auto-
From such a standpoint, it is also important to develop an educational matic braking had little effect on whether such systems made them feel
method that allows drivers to understand the functions of automation safer or made them inattentive, or on whether they believed all cars should
systems correctly, including how to disable them. be equipped with them.

6. Conclusion
Conflicts of interest
In this study, we have investigated the public misunderstanding of the
functionality and limitations of automatic braking systems. Our results The authors declare no conflict of interest.

Appendix A. Functionality of automatic braking systems from major automotive manufacturers

Manufacturer: product Detection devices Compatible obstacles Operating speeds Functionality


name

Ford: active city stop™ Radar Preceding vehicle ≤50 km/h Reduce speed
GM: front automatic Proprietary Preceding vehicle ≥4 km/h Warning, vibration,
braking™ (Engages within 60 m of preceding speed reduction,
vehicle) complete stopping
Honda: collision Combined Preceding vehicle, oncoming vehicles, pedestrians ≥5 km/h (preceding vehicle) Sound, display, speed
mitigation braking millimeter-wave radar ⁎Does not detect motorcycles or bicycles 5–80 km/h (oncoming vehicles, reduction, complete
system™ (CMBS) and monocular camera pedestrians) stopping
Honda: city-brake active Proprietary Automobiles 5–30 km/h Speed reduction,
a
system™ Does not detect motorcycles or bicycles complete stopping
Mazda: smart city brake Camera Preceding vehicle, pedestrians 4–80 km/h (vehicles) Audio warning, speed
support™ (SCBS) 10–80 km/h (pedestrians) reduction
Mazda: smart brake Combined camera and Preceding vehicle ≥15 km/h Alarm, speed
support™ (SBS) millimeter-wave radar reduction
Mercedes-Benz: active Proprietary Preceding vehicle, obstacles 7–50 km/h Warning light,
brake assist™ warning sound, speed
reduction
Mercedes-Benz: Short- and Preceding vehicle, vehicles merging ahead, vehicles 7–200 km/h Warning light,
PRE-SAFE Brake™ medium-range radar, crossing path ahead, pedestrians, obstacles on road 7.2–72 km/h (‘sudden obstacle’ warning sound, speed
stereo camera detection) reduction
≤50 km/h (collision avoidance)
Mercedes-Benz: Proprietary Preceding vehicle, obstacles 7–105 km/h (preceding vehicle) Speed reduction
collision prevention 7–50 km/h (obstacles)
assist plus™ (CPA
Plus)
Nissan: forward Camera Automobiles, pedestrians 10–80 km/h Buzzer, display, speed
emergency braking™ ≤30 km/h (collision avoidance) reduction, complete
stopping
Subaru: EyeSight™ Stereo camera Automobiles (including those coming Collision avoidance at speed differentials HUD alert, speed
from the side of car) of ≤30 km/h with respect to obstacle reduction, complete
Bicycles (including “jutting out” in path of car) Damage mitigation at higher differentials stopping
Pedestrians (including those crossing path of car) Collision avoidance at speed differentials
of ≤50 km/h with respect to preceding
vehicle
Damage mitigation at higher differentials
Toyota: pre-crash safety Combined monocular Preceding vehicle, pedestrians ≥10 km/h (preceding vehicle) Buzzer, display, speed
system™ camera and 10–80 km/h (pedestrians) reduction, complete
millimeter-wave radar Collision avoidance at speed differentials stopping
of ≤40 km/h (preceding vehicle) and ≤30
km/h (pedestrians)
Volvo: city safety™ Camera, Pedestrians, bicycles (including “sudden appearance” Proprietary Warning, speed
millimeter-wave radar from the side), preceding vehicle, oncoming traffic reduction
when turning across lanes

Collected in September 2016 from details listed on each manufacturer's Japanese-language homepage.
a
Functionality does not include complete stops if “complete stopping” is not explicitly written.
K. Shimazaki et al. / IATSS Research 42 (2018) 221–229 229

References [7] B. Schoettle, M. Sivak, Public Opinion about Self-Driving Vehicles in China, India,
Japan, the U.S., the U.K., and AustraliaMichigan, USA http://deepblue.lib.umich.
[1] National Highway Traffic Safety Administration (NHTSA), Federal Automated Vehi- edu/bitstream/handle/2027.42/109433/103139.pdf?sequence=1N 2014, Accessed
cles Policy, Accelerating the Next Revolution in Roadway Safety, https://www.nhtsa. date: 30 September 2016.
gov/sites/nhtsa.dot.gov/files/federal_automated_vehicles_policy.pdf 2016, Accessed [8] R. Cyganski, E. Fraedrich, B. Lenz, Travel-time valuation for automated driving: a
date: 26 May 2017. use-case-driven study, Transportation Research Board 94th Annual Meeting, 2015.
[2] Federal Aviation Authority Human Factors Team, The Interfaces between Flight [9] M. Kyriakidis, R. Happee, J. Winter, Public opinion on automated driving: results of
Crews and Modern Flight Deck Systems, http://www.tc.faa.gov/its/worldpac/ an international questionnaire among 5000 respondents, Transport. Res. F: Traffic
techrpt/hffaces.pdf 1996, Accessed date: 5 May 2017. Psychol. Behav. 32 (2015) 127–140.
[3] E. Bekiaris, S. Petica, V. Vicens, V. Portouli, C. Papakonstantinou, B. Peters, P. Artaud, [10] M. Wahlström, How to Study Public Imagination of Autonomous Systems: The Case
et al., SAVE system for effective assessment of the driver state and vehicle control in of the Helsinki Automated Metro, AI & Society, 2017https://doi.org/10.1007/
emergency situations – driver needs and public acceptance of emergency control s00146-017-0689-4.
aids. SAVE Consortium, 1–278, http://cordis.europa.eu/pub/telematics/docs/tap_ [11] K. Sommer, Continental Mobility Study 2013, 2013.
transport/save_d3.1.pdfN 1996, Accessed date: 30 September 2016. [12] Cabinet Office, White Paper on Traffic Accidents and Traffic Safety Measures in 2013,
[4] D. Howard, D. Dai, Public Perceptions of Self-Driving Cars: The Case of Berkeley, http://www8.cao.go.jp/koutu/taisaku/h26kou_haku/zenbun/genkyo/h1/h1b1s2_3.
CaliforniaPaper presented at the 93rd Annual Meeting TRB, Washington, DC htmlN 2014, Accessed date: 30 September 2016.
http://www.danielledai.com/academic/howard-dai-selfdrivingcars.pdfN 2014 [13] S.V. Casley, A.S. Jardim, A.M. Quartulli, A Study of Public Acceptance of Autonomous
Accessed September 30, 2016. Cars (Bachelor of Science), Worcester Polytechnic Institute, Worcester, MA, USA,
[5] W. Payre, J. Cestac, P. Delhomme, Intention to use a fully automated car: attitudes 2013.
and a priori acceptability, Transport. Res. F: Traffic Psychol. Behav. 27 (2014) [14] J.D. Power, 2012 U.S. Automotive Emerging Technologies Study Results, 2012.
252–263. [15] J. Missel, Ipsos MORI Loyalty Automotive Survey, 2014.
[6] B. Schoettle, M. Sivak, A Survey of Public Opinion about Autonomous and Self-Driv- [16] TE Connectivity, TE Connectivity Survey Finds Safety the Top Consumer Priority in
ing Vehicles in the U.S., the U.K., and AustraliaMichigan, USA http://deepblue.lib. Adopting Autonomous Vehicle Technology, 2013.
umich.edu/bitstream/handle/2027.42/108384/103024.pdfN 2014, Accessed date:
30 September 2016.

You might also like