Content uploaded by Hillary Abraham
Author content
All content in this area was uploaded by Hillary Abraham on Sep 28, 2017
Content may be subject to copyright.
What’s'in'a'Name:'Vehicle'Technology'Branding'&'
Consumer'Expectations'for'Automation'
Hillary Abraham
MIT AgeLab
Cambridge, US
habraham@mit.edu
Bobbie Seppelt
MIT AgeLab &
Touchstone
Evaluations
Cambridge, US
bseppelt@mit.edu
Bruce Mehler
MIT AgeLab
Cambridge, US
bmehler@mit.edu
Bryan Reimer
MIT AgeLab
Cambridge, US
reimer@mit.edu
ABSTRACT'
Vehicle technology naming has the potential to influence
drivers’ expectations (mental model) of the level of
autonomous operation supported by semi-automated
technologies that are rapidly becoming available in new
vehicles. If divergence exists between expectations and
actual design specifications, it may make it harder to
develop trust or clear expectations of systems, thus
mitigating potential benefits. Alternately, over-trust and
misuse due to misunderstanding increase the potential for
adverse events. An online survey investigated whether and
how names of advanced driver assistance systems (ADAS)
and automation features relate to expected automation
levels. Systems with “Cruise” in their names were
associated with lower levels of automation. “Assist”
systems appeared to create confusion between whether the
driver is assisting the system or vice versa. Survey findings
indicate the importance of vehicle technology naming and
its impact in influencing drivers’ expectations of
responsibility between the driver and system in who
performs individual driving functions.
Author'Keywords'
Advanced Driver Assistance Systems; Branding;
Automation; Confusion
CCS'Concepts'
• Human-centered computing~User centered design
INTRODUCTION'
Most automotive manufacturers now offer, or are currently
pursuing research on, advanced driver assistance systems
(ADAS) and automated driving features. Collectively,
semi-automated vehicle technologies (ADAS and lower
level automation systems) are rapidly becoming standard or
optional features on new vehicles. In order to help provide
common definitions for different types of automation in
vehicles, the Society of Automotive Engineers (SAE)
developed a taxonomy with detailed descriptions for
vehicles equipped with automated features [24]. At present,
consumers are only able to purchase vehicles equipped with
driver assistance (Level 1) and partial automation (Level 2)
systems. However, several automotive manufacturers have
announced production vehicles to be available this year
with conditional automation (Level 3). High automation
(Level 4) technologies are being tested globally with
expected commercial availability being forecast in less than
5 years [15].
Efforts to develop ADAS and automation features are based
upon manufacturer-specific design specifications. These
specifications aim to produce a technology with the
capability to perform in a particular operational design
domain (ODD). The system implementation and specific
use conditions encompassed in the static and dynamic
aspects of the ODD [28] are representative of a system
designer’s mental model for the technology. How drivers
learn about individual systems is influenced by their pre-
existent mental models – those formed prior to initial use,
e.g., from exposure to other technologies [12]. A driver’s
mental model aids him or her in understanding a system’s
ODD, interface characteristics and other system limitations
necessary for proper system use [4,27]. While driver
education and other more active methods for encouraging
proper use (in vehicle coaching, etc.) face challenges at
each level of automation, the most relevant current
challenge exists with partial driving automation (Level 2),
for which governments, businesses, researchers and
consumers have argued the marketing name of a system
may promote the misalignment of driver and designer
expectations [5,7,18]. In Level 2 automation, the system
performs sustained lateral and longitudinal management of
the driving task, while the driver performs the remaining
subtasks, including object and event detection and response
(OEDR). Driver belief that a system has the ability to
perform OEDR at a level greater than the system’s design
characteristics may lead to misuse [22].
Human Machine Interfaces (HMIs) for automated features
are intended, by design, to help support driver
understanding of features and to promote proper system
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page. Copyrights for
components of this work owned by others than ACM must be honored.
Abstracting with credit is permitted. To copy otherwise, or republish, to
post on servers or to redistribute to lists, requires prior specific permission
and/or a fee. Request permissions from Permissions@acm.org.
AutomotiveUI '17, September 24–27, 2017, Oldenburg, Germany
© 2017 Association for Computing Machinery.
ACM ISBN 978-1-4503-5150-8/17/09…$15.00
https://doi.org/10.1145/3122986.3123018
use. At Levels 1 and 2, in which features assist drivers for
only a partial set of the dynamic tasks of driving, their
HMIs aim to support drivers in maintaining their attention
to the roadway. One adopted implementation strategy to
support this aim (e.g. Tesla, Volvo, etc.) is to require
drivers to keep their hands on the wheel with minimal
steering input; however, the amount of input and amount of
time a driver can go before hands-off-wheel warnings are
issued varies between system and use conditions, resulting
in the potential for prolonged intervals of declining
situation awareness. Further, there is not currently a proven
link between hands-on-wheel during Level 2 use and
situational awareness. Looking to enforce a greater degree
of control on driver attentiveness, GM’s SuperCruise,
anticipated to be commercially available in the 2018
Cadillac CT6, is reported to be designed with an integrated
head pose detection system in order to monitor driver
awareness and to trigger a range of cues to promote driver
attentiveness [8]. The standardization of such approaches is
currently under consideration in Europe [9] and is
supported by research [23].
Multiple factors contribute to a driver’s expectations of
system capability [e.g., 1,13,21,22,26]. Drivers’ attitudes
and beliefs about system capability and performance are
known to influence their use of technology [6,10,14,30].
Factors such as a driver’s prior experience with similar
technologies, predisposed trusting tendencies, and attitudes
formed from exposure to media and societal opinion might
all contribute to a driver’s belief that a system can handle a
task outside of its ODD.
The name of a driver assistance system also has the
potential to impact their perceptions of system capability.
From consumer psychology research, there is an ascribed
importance of branding and the names assigned to products;
naming influences expectation of product attributes and
preconditions consumers to assign valence based on
induced biases [17]. In application to driving automation
systems, the names assigned to technologies have the
potential to shape driver perceptions in a way that bias
attitudes and affect use [30]. Other than a small survey by
Tesla [31], little structured research has investigated
whether the name of a system impacts driver expectations
of a system, particularly in relation to what the driver
expects their role should be while using the vehicle and
system.
As brand names are increasingly used to discuss vehicle
automation systems with a vast range of design models,
improved understanding of whether or not brand names of
current and proposed driver assistance / automation systems
impact driver expectations may help guide future naming
discussions and considerations for standardization. A
survey was designed to investigate two primary research
questions:
1.!Does the name of driver assistance systems affect
a customer’s perception of the level of automation
of that system?
2.!If so, do commonly used terms when branding
ADAS (e.g. Auto, Pilot, Assist, Cruise) direct
consumer perceptions toward presumptions of
lower or higher levels of automation?
METHOD'
Participants'
Participants were recruited using online advertisements and
web posts to the MIT AgeLab website. In total, a
convenience sample of 453 participants was analyzed. The
sample was 37% male and 61% female; the remaining 2.6%
of individuals selected “Other or choose not to answer.”
Age of respondents ranged from 20-69, with 30% of
respondents in their 20s, 19% in their 30s, 6% in their 40s,
18% in their 50s, and 27% in their 60s. Respondents were
generally highly educated; 38% had completed a graduate
or professional degree as their highest level of education,
18% had completed some graduate education, 29% had
completed a Bachelor’s degree, 2% had an Associate’s
degree, 1% had a trade school certificate, 12% had
completed some college, 1% had graduated high school,
and 0% had completed some high school. Most respondents
(71%) were from the state of Massachusetts in the USA.
Survey'Instrument'
Systems'Addressed'
Nineteen driver assistance systems were selected for
inclusion in the survey (Table 1). Attempts were made to
incorporate all systems commercially available or publicly
proposed at the time of survey deployment that feature both
adaptive cruise control and a lane centering component, yet
require the driver to engage in some of the dynamic aspects
of driving, either actively or as a fallback-ready user (e.g.
Level 1 – Level 3). Researchers were particularly interested
in how common English terms might affect perceptions of
system capabilities; as such, systems that included the name
of the manufacturer in their title were not included (e.g.
Honda Sensing). Four fabricated system names were
included in the survey to explore differences between terms
typically used in systems at higher levels of automation and
those typically used for systems at lower levels.
Automation'Categories'
Seven descriptions of differing levels of automation were
created for participants to classify systems (Figure 1).
These categories were developed based on the six SAE
J3016 levels of automation [24], plus an additional level
(“L1.5,” conceptually between 1 & 2) to accommodate
commercially available systems that require the driver to
keep their hands on the wheel at certain frequencies, as a
function of the adopted implementation strategy, in order to
perform continuous lane centering. Care was taken to
ensure these categories accurately represented J3016 levels,
while simultaneously being understandable to the layman in
terms of the division of driving task responsibility.
Particular attention was paid to the distinction between
general tasks the driver would be responsible for, versus
general tasks the driving assistance system would be
responsible for, while the system was engaged or active.
Categorizations generalized ODD and dynamic driving task
(DDT) into broad categories of responsibility, rather than
listing and requesting classifications for individual ODDs
and DDTs, in an attempt to avoid overwhelming the survey
respondents (Figure 1).
Survey'Design'Methodology'
After selecting systems for inclusion and developing a first
draft of automation categories, a survey instrument was
developed by the research team. This instrument went
through a series of internal revisions before piloting with
additional staff members not involved in the project to
ensure layman understanding of all terms and definitions
involved. After piloting, research staff spoke informally
with pilot subjects about the survey design, format, and
clarity of questions. Pilot subject feedback was integrated
into the final instrument detailed within this report.
Survey'Procedure'
Participants were first presented with a brief introduction to
the survey and a description of each level of automation
(Figure 1). After reading the introduction and level
classifications, participants were asked to imagine they
were the driver in a vehicle equipped with an automated
system. Participants were then provided with the list of 19
systems. For each system, participants selected the category
from the seven levels of automation that best described the
division of task responsibility that they would expect to
exist between them as the driver and the system. In order to
maximize the likelihood that categorization would be made
based on name alone, survey takers were instructed not to
use any outside resources when making their categorization.
After assigning a level to a system, participants rated their
confidence in their level assignment on a 5-pt scale ranging
from 1 (low confidence) to 5 (high confidence). This was
repeated for all 19 systems.
After assigning every system to a category of automation
and rating their confidence in their assignment, participants
were asked, “before taking this survey, how familiar were
you with any of the systems?” and provided a 5-pt scale
ranging from “Not familiar at all” to “Extremely familiar.”
Participants were asked six questions to gauge their early
adopter status, vehicle information, and whether or not any
of their vehicles had any of the survey systems installed.
System
Manufacturer
Availability
LoA
Active Cruise Control
BMW
Available
1
AutoCruise
N/A
N/A
N/A
Autopilot
Tesla
Available
2
Distronic Plus
Mercedes-Benz
Available
1
Drive Pilot
Mercedes-Benz
Available
1.5
Driving Assistant Plus
BMW
Available
1.5
Enhanced Autopilot
Tesla
In Development
3
Eyesight
Subaru
Available
1
Highway Pilot
Audi
In Development
3
Intelligent Assist
N/A
N/A
N/A
Intelligent Cruise Control
Nissan
Available
1
Intelligent Drive
Mercedes-Benz
Available
1
Intelligent Pilot
N/A
N/A
N/A
Pilot Assist
Volvo
Available
1.5
Pilot Plus
N/A
N/A
N/A
ProPilot
Nissan
In Development
1.5
Super Cruise
GM
In Development
2
Traffic Jam Assist
Audi
Available
1.5
Traffic Jam Pilot
Audi
In Development
3
Table 1. Systems, availability, and level of automation (LoA)
at time of survey deployment.
Figure 1. After a survey introduction, participants were presented with this graphic representing seven categories of automation,
ranging from SAE Level 0 (fully manual, far left box) to SAE Level 5 (fully automated, far right box). These seven categories
provide layman’s definitions of the division of driving task responsibility between driver and system.
The survey ended by collecting demographic information,
including date of birth, highest level of education,
employment status, household income, gender, and zip
code.
Participants who completed the survey were offered the
opportunity to enter a raffle to win one of 10 $50 Amazon
gift cards. The survey was constructed in Qualtrics, and
participants were asked to take the survey online via a
desktop or laptop computer. The survey was open for data
collection from February 22nd – March 6th 2017.
RESULTS'
Data were analyzed using SPSS Version 24. As analyses
were run multiple times (once for each system), a
Bonferroni correction was used to determine significance.
Significance was set at p < .0026 for analyses of all 19
systems (.05 / 19), and p < .0033 for analyses of the 15
deployed or in-development systems (.05 / 15). For age
analyses, respondents were grouped into five age ranges:
20-29, 30-39, 40-49, 50-59, and 60-69. '
Familiarity'&'Correctness'
Most participants selected “not familiar at all” for
familiarity with each of the systems prior to taking the
survey (Figure 2). Two systems, Active Cruise Control and
Autopilot, had higher levels of familiarity than the other
systems in the sample. Importantly, it is unclear whether or
not respondents were familiar with Tesla’s Autopilot, the
term “autopilot” within the context of aviation, or the
colloquial “autopilot,” used when referring to completing a
task absentmindedly or without focus. While more
respondents were familiar with these systems, more than
half (54.5% and 66.2% respectively) reported being either
“not familiar at all” or only “slightly familiar” with either
system.
Table 2. Overall accuracy for system categorization was low. There was no relationship between correct categorization and
confidence. Most participants did not select L0 for most systems.
Figure 2. Participants rated themselves as being not at all familiar with most of the systems prior to taking the survey.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Act iv e
Cruis e
Control
Autopilot Distronic
Plus
Dri ve Pi lo t Driving
Assistant
Plus
Enh an ced
Autopilot
Eyesight Hi g hway
Pilot
Intelligent
Cruis e
Control
Intelligent
Dri v e
Pilo t Assist ProPil ot Supercru ise
Traf fi c J am
Assist
Traffic
Jam Pilot
Not at al l fam il ia r
Slig htly familiar
Moderat ely familiar
Very fa mi li ar
Ext re mel y f ami li ar
Most respondents did not accurately classify most systems
into their correct level of automation. While most systems
had a slightly higher percentage of correct categorizations
than would be expected from random guessing (14%
correct), overall accuracy remained low (Table 2). Active
Cruise Control had the highest proportion of correct
categorizations, with 50% of respondents correctly
categorizing it as a Level 1 system. Intelligent Drive, also a
Level 1 system, had the lowest proportion of correct
categorizations (9%).
Confidence ratings varied across systems. On a scale of 1
(low confidence) to 5 (high confidence), over half of
participants rated their confidence as a 4 or 5 for Active
Cruise Control, AutoCruise, and Autopilot. Distronic Plus
had the highest proportion of low confidence ratings, with
66% of participants rating their confidence at a 1 or 2.
The Mann-Whitney U test was applied to investigate
differences in confidence of system categorization between
respondents who correctly versus incorrectly classified a
system. Most systems showed no significant difference in
confidence rating between individuals that correctly
categorized a system compared to those that incorrectly
categorized the system (Table 2).
One system (EyeSight) showed significant age differences
in correct categorization; no other system showed
significant differences between different age groups and
correct categorization (Table 3). There were no significant
gender differences in correct categorization of any system;
men were not more frequently correct than women and vice
versa. However, there were significant differences in
confidence and familiarity ratings between genders (Table
3). In all significant cases, men rated themselves more
confident than women in their responses. Men also reported
being more familiar with systems prior to taking the survey.
Categorization'Patterns'
Chi-square goodness of fit tests were used to explore
whether or not the distribution of level categorization
differed from random guessing; that is, equal distribution of
responses across the seven levels of automation. Chi-
squared residuals were explored to determine which cells
contributed most toward the results [29].
Every system showed a significant difference from equal
distribution of categorizations (Table 2). When exploring
the raw residuals, one system (EyeSight) showed residuals
close to random guessing. The remaining 18 systems
showed that a disproportionate number of participants were
not selecting L0 (no automation) for most systems. Aside
from EyeSight, every system had an L0 residual of less than
-25, and ten had residuals of less than -50. As the survey
instructions indicated each system was an automated
system, it seems plausible that respondents were not
selecting L0 due to the survey design, rather than any
impact of naming. As such, chi-square analysis was re-run
using only the data points assigned to L1-5. Each system
still showed a significant difference from equal distribution
of categorizations (Table 4).
The residuals for the second set of chi-square analyses
revealed two strong relationships between name and
categorization within the 19 systems (Table 4). Table 4 was
color-coded to more easily display patterns in the residuals.
Dark green cells are those with the highest residuals, dark
grey are those with the lowest, and white are those with
Table 3 No significant differences were exhibited in gender and accuracy, but significant gender differences were exhibited in
confidence and familiarity with systems.
residuals closest to zero. The first relationship was between
the systems with “cruise” in the name; these four systems
were consistently rated at the lower end of the automation
scale. They also generally received higher confidence
ratings. The second relationship was regarding the four
“assist” systems, which received high frequency of
categorizations in L1.5 and L3. Confidence ratings were
consistently in the middle for these systems. Four systems
(ProPilot, Highway Pilot, Distronic Plus, and EyeSight)
showed residuals that were widely distributed across the
automation scale; these systems were still significantly
different from random guessing, but had the lowest range of
residuals. They also showed the lowest confidence ratings
of all systems. The remaining 8 systems showed responses
different from guessing, but no easily identified pattern was
apparent between any of the 8 systems and their names.
An alternate visualization of the results appears in Figure 3.
Here, colored squares represent mean response, while black
lines indicate the bounds of the first and third quartiles.
While the mean categorization for most systems is higher
than the correct categorization, it is important to note that
mean has limited value for interpretation for two reasons:
first, the categories provided are ordinal with dissimilar
differences between each category. Second, as some
participants were likely guessing, there were more
opportunities to select levels of automation above the
correct category than below. The quartiles bounds indicate
some systems, such as Active Cruise Control, Intelligent
Cruise Control, and AutoCruise, have short distributions
ranging between L1 and L2. Others, such as Highway Pilot,
Traffic Jam Pilot, and Intelligent Assist, have wider
distributions spanning from L1.5 to L4.
DISCUSSION'
The first research question was, does the name of driver
assistance systems affect a customer’s perception of the
level of automation of that system. The survey results
indicate that the name of a system does have an impact on
the degree of responsibility that respondents expected to
have as a driver when using a partially automated system.
Overall reported familiarity with the systems was low and
participants were instructed not to use outside resources
when categorizing systems. Consequently, the primary
information contributing to significantly different
categorizations was likely centered on the name of the
system. Initial exploration into age effects suggests these
results are pervasive across all ages, though small sample
sizes for respondents in their 40s may limit interpretation of
these results.
Active Cruise Control was the only system that a majority
of respondents categorized correctly, and it received the
highest confidence and familiarity ratings. Tesla’s
Autopilot also received comparatively high familiarity
ratings, but accuracy was in line with other systems. Low
accuracy could be due to Autopilot’s name, but as prior
familiarity with the system was notable, little can be said
about the effects of solely the term “Autopilot” on
determination of system capability. Rather, the differing
results of these two higher familiarity systems suggest
Different from
Guessing, no L0
Residuals
Confidence**
System
X2
df
p
L1
L1.5
L2
L3
L4
L5
Median
Mode
"Cruise"
systems:
lower
levels
Active Cruise Control
473.9
5
<0.001
159
5
-41
-20
-53
-50
4
5
AutoCruise
175.3
5
<0.001
65
60
-9
-18
-51
-45
4
4
Intelligent Cruise Control
149.9
5
<0.001
63
52
-6
-30
-42
-39
3
4
Super Cruise
68.2
5
<0.001
-22
40
38
-4
-17
-34
3
1
"Assist"
systems:
split
between
1.5 & 3
Driving Assistant Plus
139.6
5
<0.001
-45
38
13
61
--18
-51
3
3
Intelligent Assist
38.4
5
<0.001
-23
35
-9
23
-4
-20
3
3
Pilot Assist
156.5
5
<0.001
-35
66
17
39
-32
-56
3
3
Traffic Jam Assist
70.0
5
<0.001
1
39
-14
34
-17
-43
3
3
Closest to
random
guessing
ProPilot
59.5
5
<0.001
-50
-22
20
20
20
13
3
1
Highway Pilot
67.6
5
<0.001
-42
10
24
33
10
-36
3
3
Traffic Jam Pilot
32.4
5
<0.001
-26
15
10
27
-1
-24
3
3
Distronic Plus
48.5
5
<0.001
-24
15
30
24
-24
-23
1
1
EyeSight*
36.8
5
<0.001
-11
23
-6
32
-20
-16
3
1
Drive Pilot
89.7
5
<0.001
-46
10
39
38
-3
-36
3
1
Pilot Plus
113.6
5
<0.001
-51
2
46
23
30
-49
3
1
Enhanced Autopilot
97.6
5
<0.001
-59
-28
-5
42
30
20
3
3
Intelligent Pilot
100.7
5
<0.001
-54
-2
41
46
-5
-25
3
3
Intelligent Drive
76.6
5
<0.001
-31
48
31
5
-25
-27
3
3
Autopilot
58.6
5
<0.001
-47
-2
43
14
-2
-5
3
4
*Eyesight also showed a large proportion of responses on L0
**Confidence was rated on a 5-pt scale, with 1 being low confidence & 5 being high confidence
Table 4. Raw residuals of Chi-Square Goodness of Fit Tests for equal distribution of responses across Levels 1 – 5 showed three
patterns in name and level categorization.
outside factors (e.g., hands-on experience, media reports,
educational material) likely have a greater impact than
name alone on setting expectations for a system.
Setting expectations from the beginning is important, and
first impressions have a long been known to influence use
[19]. Nevertheless, misconceptions in perceptions of a
system can be overridden as a consumer receives more
information and first-hand experience with a system.
Moving forward, manufacturers or other parties will need to
continue investing in appropriate ways of educating drivers
on responsible use of their partially automated driving
system. As one example of a more integrated education
approach, Subaru has developed asales and delivery system
for the EyeSight system that aims to help enhance
consumer understanding throughout the purchase process
[1]. While Subaru’s developments in this area may be
industry leading, other manufactures have the opportunity
to leverage observations surrounding consumers’ use of
current systems [28] to inform and refine models for use
during the sales process or real-time coaching.
The second research question was, do commonly used
terms when branding ADAS (e.g. Auto, Pilot, Assist,
Cruise) direct consumer perceptions toward presumptions
of lower or higher levels of automation? Survey results
indicated that terms, to varying degrees, influence
consumers’ perceptions of automation level. For example,
Cruise Control is an established in-vehicle technology with
which many drivers are familiar. Unsurprisingly, “cruise”
systems were frequently rated at the lower end of the
automation scale. It appears drivers interpreted “cruise”
systems to be slightly more automated than cruise control,
setting an expectation that while a “cruise” system might be
able to handle part of the driving task, ultimate
responsibility remained on the driver. Though name may
not be the most important factor for setting consumer
expectations, manufacturers could benefit from leveraging
understood terminology from established and high-
familiarity systems to properly orient consumer
understanding of their responsibilities while driving and
using a system.
Care should be taken when using potentially ambiguous
terms to name systems. “Assist” systems, which attempt to
indicate that the driver should not be relying on the vehicle
to complete all of the driving tasks, were confused between
two non-adjacent level classifications. Participants
frequently either rated the system as L1.5, which involves
providing speed and steering support while requiring the
driver to keep their hands on the wheel, or L3, which
involves the system handling most tasks and the driver
being prepared to take over if requested. In one (L1.5), the
system is assisting the driver, who holds responsibility for
OEDR. In the other (L3), the driver is expected to be ready
to assist the system, which is responsible for the dynamic
driving tasks. These two levels require very different input
from the driver, and avoiding confusion regarding
his/herrole is crucial.
Overall, the results suggest that brand names do influence
perceptions of technologies; yet, brand names do not offer
enough information to appropriately set driver expectations
for their role while driving. The wide distribution of
responses for some systems, notably Highway Pilot,
Traffic Jam Pilot, and Intelligent Assist, indicate that name
of a system may be interpreted numerous different ways by
individual consumers. As many consumers own more than
one vehicle, a greater degree of commonality of design and
naming characteristics (e.g. ABS, ESC, etc.) in combination
with increased driver education may be critical in the
successful transformation of personal mobility from largely
manual control through to higher levels of automation. As
the aviation literature has long stipulated, with increasing
Figure 3. Simplified visualization of level classification distributions. Colored boxes indicate mean ranking, and black lines
represent bounds of the first and third quartiles.
levels of automation, increased education is required to
ensure operators are fully aware of their role [26]. It is
unclear how current automotive and technology developers
building automated driving systems across the levels are
fully embracing human-centered design principles from
conceptualization to technology naming, marketing,
delivery, and eventual use. It is clear that naming
conventions could be used to help amplify system
intuitiveness (e.g., where the driver’s and system’s mental
models by nature have a high degree of overlap), and to
better facilitate adoption of automated driving features that
have the potential to revolutionize mobility and increase
vehicle safety.
Consistent with previous gender research, men in this
survey were more confident in their categorizations when
they were in fact incorrect than were women who were
incorrect [16]. This overconfidence, combined with a
general male preference to seek out information on their
own rather than to be provided with assistance from
dealership staff or the car itself [3], could create additional
challenges for male consumers in setting appropriate driver
mental models at first exposure to a system. As these results
suggest, name alone is not enough to appropriately orient
drivers to system limitations and appropriate use.One
solution might be to necessitate a self-guided tutorial or
other training sessions run by the system itself, in which the
system could be locked until the driver completes the
tutorial.
As research on this topic expands, it is recommended that a
naming guide be created to provide insight for
manufacturers when designing and marketing new systems.
To that end, future research on this topic would benefit
from a larger, more nationally representative sample. This
survey was limited in the age ranges represented in the
sample, as well as the geographic distribution and education
level of respondents. Future research may also need to limit
the number of systems addressed. Due to the similarity and
overlapping nature of many of the technology names (e.g.,
Traffic Jam Pilot, Traffic Jam Assist, and Pilot Assist), it
was difficult to interpret which term was affecting
classification to the higher degree. A more targeted
approach could provide deeper insight toward the specific
portions of each name and how they relate to automation
and driver role expectations.
ACKNOWLEDGMENTS'
Support for this work was provided by the Advanced
Vehicle Technology (AVT) consortium at MIT. The views
and conclusions being expressed are those of the authors,
and have not been sponsored, approved, or endorsed by
members of the consortium.
REFERENCES'
1.!G. Abe and J. Richardson. 2006. Alarm timing, trust
and driver expectation for forward collision warning!
systems. Applied Ergonomics, 37(5), 577-586.
2.!Hillary Abraham, Hale McAnulty, Bruce Mehler, and
Bryan Reimer. 2017. A Case Study of Today’s
Automotove Dealerships: The Introduction and
Delivery of Advanced Driver Assistance Systems. In
Transportation Research Record: Journal of the
Transportation Research Board 2660. DOI:
10.3141/2660-02.
3.!Hillary Abraham, Chaiwoo Lee, Samantha Brady,
Craig Fitzgerald, Bruce Mehler, Joseph F. Coughlin.
2017. Autonomous Vehicles and Alternatives to
Driving: Trust, Preferences, and Effects of Age. In
Proceedings of the Transportation Research Board 96th
Annual Meeting (TRB ’17).
4.!M. Beggiato and J. F. Krems. 2013. The evolution of
mental model, trust and acceptance of adaptive cruise
control in relation to initial information. Transportation
research
5.!Neal E. Boudette. 2017. Tesla’s Self-Driving System
Cleared in Deadly Crash. Retrieved May 4, 2017 from
https://www.nytimes.com/2017/01/19/business/tesla-
model-s-autopilot-fatal-crash.html
6.!J. K. Choi and Y. G. Ji. 2015. Investigating the
importance of trust on adopting an autonomous
vehicle. International Journal of Human-Computer
Interaction, 31(10), 692-702.
7.!Consumer Reports. 2016. Tesla’s Autopilot: Too Much
Autonomy Too Soon. Retrieved April 4, 2017 from
http://www.consumerreports.org/tesla/tesla-autopilot-
too-much-autonomy-too-soon/
8.!Alex Davies. 2017. Cadillac Cracks a Self-Driving
Puzzle by Shoving a Camera in your Face. Retrieved
May 15, 2017 from
https://www.wired.com/2017/04/cadillac-cracks-self-
driving-puzzle-shoving-camera-face/
9.!European Commission. 2016. Status of the review of
the General Safety and Pedestrian Safety Regulations.
Presentation. (14 December, 2016). Retrieved May 16,
2017 from
https://www.unece.org/fileadmin/DAM/trans/doc/2016
/wp29grsp/GRSP-60-21e.pdf
10.!M. Ghazizadeh, Y. Peng, J. D. Lee, and L. N. Boyle.
2012. Augmenting the technology acceptance model
with trust: Commercial drivers’ attitudes towards
monitoring and feedback. In Proceedings of the Human
Factors and Ergonomics Society Annual Meeting (Vol.
56, No. 1, pp. 2286-2290). Sage CA: Los Angeles, CA:
Sage Publications.
11.!J. L. Harbluk, Y. I. Noy, and M. Eizenman. 2002. The
impact of cognitive distraction on driver visual
behaviour and vehicle control (No. TP# 13889 E).
12.!T. A. Kazi, N. A. Stanton, G. H. Walker, and M. S.
Young, M. S. 2007. Designer driving: drivers'
conceptual models and level of trust in adaptive cruise
control. International journal of vehicle design, 45(3),
339-360.
13.!J. D. Lee and N. Moray. 1994. Trust, self-confidence,
and operators' adaptation to automation. International
Journal of Human-Computer Studies, 40, 153-184.
14.!J. D. Lee and K. A. See. 2004. Trust in automation:
Designing for appropriate reliance. Human
factors, 46(1), 50-80.
15.!Paul Lewis, Gregory Rogers, and Stanford Turner.
2017. Beyond Speculation: Automated Vehicles and
Public Policy. Retrieved May 16, 2017 from
https://www.enotrans.org/wp-
content/uploads/2017/04/AV_FINAL.pdf
16.!Mary Lundenberg, Paul W. Fox, and Judith Punccohar.
1994. Highly Confident but Wrong: Gender
Differences and Similarities in Confidence Judgments.
Journal of Educational Psychology, 86, 114-121.
17.!D. Maheswaran, D. M. Mackie, and S. Chaiken. 1992.
Brand name as a heuristic cue: The effects of task
importance and expectancy confirmation on consumer
judgments. Journal of consumer psychology, 1(4), 317-
336.
18.!Russ Mitchell. 2016. Controversy over Tesla
“autopilot” name keeps growing. Retrieved May 12,
2017 from http://www.latimes.com/business/autos/la-
fi-hy-autopilot-controversy-20160721-snap-story.html
19.!Michael G. Morris and Andrew Dillon. 1997. How user
perceptions influence software use. IEEE software, 14,
4, 58-65.
20.!NHTSA. 2017. Investigation PE 16-007.
21.!D.A. Norman. 1990. The 'problem' with automation:
Inappropriate feedback and interaction, not 'over-
automation'. Philosophical Transactions of the Royal
Society London, Series B, Biological Sciences,
327(1241), 585-593.
22.!R. Parasuraman, R. and V. Riley. 1997. Humans and
automation: Use, misuse, disuse, abuse. Human
Factors: The Journal of the Human Factors and
Ergonomics Society, 39(2), 230-253.
23.!Bryan Reimer, Anthony Pettinato, Lex Fridman,
Joonbum Lee, Bruce Mehler, Bobbie Seppelt, Junghee
Park, and Karl Iagnemma. 2016. Behavioral Impact of
Drivers’ Roles in Automated Driving. In Proceedings
of the 8th International Conference on Automotive User
Interfaces and Interactive Vehicle Applications
(AutomotiveUI '16), 217-224. DOI:
10.1145/3003715.3005411
24.!SAE. 2016. Taxonomy and Definitions for Terms
Related to Driving Automation Systems for On-Road
Motor Vehicles. SAE Standard J3016
25.!S. Samuel and D. L. Fisher. 2015. Evaluation of the
Minimum Forward Roadway Glance
Duration. Transportation Research Record: Journal of
the Transportation Research Board, (2518), 9-17.
26.!N. B. Sarter, D. D. Woods, and C. E. Billings. 1997.
Automation surprises. In G. Salvendy (Ed.), Handbook
of Human Factors & Ergonomics, second edition.
Wiley.
27.!Bobbie Seppelt and L. D. Lee. 2007. Making adaptive
cruise control (ACC) limits visible. International
journal of human-computer studies, 65(3), 192-205.
28.!Bobbie Seppelt, Bryan Reimer, Linda Angell, and Sean
Seaman. 2017. Considering the Human across Levels
of Automation: Implications for Reliance. In
Proceedings of the Ninth International Driving
Symposium on Human Factors in Driver Assessment,
Training and Vehicle Design (DA ’17). In press.
29.!Donald Sharpe. 2015. Your Chi-Square Test is
Statistically Significant: Now What? Practical
Assessment, Research & Evaluation 20, 8.
30.!N. Trübswetter and K. Bengler. 2013. Why should I
use ADAS? Advanced driver assistance systems and
the elderly: knowledge, experience, and usage barriers.
In Proc. 7th Int. Driving Symposium on Human
Factors in Driver Assessment, Training and Vehicle
Design (pp. 495-501).
31.!Konrad Webner. 2016. Awareness and utilization of
the Autopilot Tesla Survey. Presentation of results:
Customer survey. Retrieved April 4, 2017 from
https://www.tesla.com/sites/default/files/blog_attachme
nts/tesla_survey_autopilot_awareness.pdf
