A Simulator Study on Car User’s Perceptions in Interaction with

Autonomous Shuttles

Sagar Meda

1 a

, Mario Ilic

2 b

, Tanja Niels

2 c

and Eftychios Papapanagiotou

1 d

Professorship for Intelligent, Multimodal Transport Systems, UniBwM, Neubiberg, Germany

Chair of Trafﬁc Engineering and Control, TUM, Munich, Germany

Keywords:

Autonomous Shuttles, Human Errors, Driving Behaviours, Perception, Safety, Stress, Driving Simulator,

Experimental Design, Qualitative Survey and Participants.

Abstract:

This study integrates a driving simulator and microscopic trafﬁc simulation tool to evaluate the impact of

autonomous shuttles on driving behavior and safety in a university campus environment. Two scenarios were

developed: one featuring a conventional shuttle and another with an autonomous shuttle, allowing a direct

comparison of driver perceptions under identical conditions. Results show that perceived safety was higher

for conventional shuttles (Mean: 5.909) compared to autonomous shuttles (Mean: 2.818), while stress levels

remained consistent across both scenarios. These ﬁndings highlight critical human factors and challenges in

integrating autonomous shuttles, offering empirical insights into their behavioral and safety implications in

mixed-trafﬁc environments.

1 INTRODUCTION

Technological advancements have signiﬁcantly im-

proved road and vehicle safety systems in recent

decades, enhancing the overall safety and driving ex-

perience. However, human errors still hold a larger

share of the causes of an accident. Accidents caused

by human error continue to be a signiﬁcant problem,

resulting in serious injuries, fatalities, and substantial

ﬁnancial losses despite trafﬁc laws and awareness ef-

forts. A study by National Highway Trafﬁc Safety

Administration (2017) found that 94% to 96% of all

vehicle accidents are due to human errors.

Human errors can result from various factors, such

as driving while distracted or intoxicated, speeding,

acting recklessly, lack of awareness and trafﬁc law vi-

olations. Accidents are more likely to occur when the

driver’s focus is distracted from the road by activi-

ties like eating, using in-car entertainment systems, or

talking on the phone. Similar to this, driving while in-

toxicated increases the likelihood of accidents by im-

pairing judgment, coordination, and response times.

Speeding is a common cause of accidents because it

impairs a driver’s capacity to react appropriately to

unforeseen circumstances. Lack of awareness and ed-

https://orcid.org/0009-0006-9227-8596

https://orcid.org/0000-0003-2457-698X

https://orcid.org/0000-0002-8530-0285

https://orcid.org/0000-0002-8538-0326

ucation due to the introduction of new mobility op-

tions such as Autonomous Vehicles (AVs) also raises

the risk and probability of accidents. A systemic ap-

proach is necessary for mitigating the issue of road

safety and accidents caused by human error. This en-

tails further research into the psychological and be-

havioural components of driving, improved educa-

tion and awareness campaigns, and enforcing stricter

regulations. Furthermore, by making up for human

error and limitations, technological developments in

vehicle technology, like advanced driver-assistance

systems (ADAS), detection sensors and autonomous

driving systems, can contribute to a safer driving en-

vironment, which potentially reduces road accidents

and their related external costs.

AVs are one such option; they have already been

driven on public roads. It has been observed that they

adhere to local laws and trafﬁc restrictions. Moreover,

a study conducted by Bartneck et al. (2021) found

that although the average human driver was seen to

be driving at 70 Miles per hour (113 Kilometers per

hour) on the Atlanta freeway, autonomous cars were

found to be travelling below the posted limits. Thus

realizing the beneﬁts hosted by such mobility options.

Further, the AVs possess superiority up to a certain ex-

tent based on the ability of the vehicle to perform any

task under varying conditions.

However, the emergence of AVs could compli-

cate the urban environment to an unprecedented de-

gree, presenting both advantages and challenges for

Meda, S., Ilic, M., Niels, T. and Papapanagiotou, E.

A Simulator Study on Car User’s Perceptions in Interaction with Autonomous Shuttles.

DOI: 10.5220/0013272700003941

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 11th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2025), pages 54-65

ISBN: 978-989-758-745-0; ISSN: 2184-495X

the modern world. The interaction between an au-

tonomous vehicle and a conventional driver can be

uncertain at times, and this might have an impact on

driving dynamics and human perception abilities to a

certain degree. Therefore, the increasing number of

cities across the world should be innovative in the in-

tegration of intelligent mobility solutions into the ex-

isting urban trafﬁc environment that would bgeable to

meet the population’s demands for efﬁcient, safe, and

sustainable transportation.

The study area considered for this study is the

campus of the University of the Bundeswehr Mu-

nich (Universit

at der Bundeswehr M

unchen), situ-

ated in Neubiberg, Munich, Germany. It spans ap-

proximately 350 acres and plans to introduce an au-

tonomous shuttle into the campus mobility ecosys-

tem. While this initiative can signiﬁcantly reduce en-

vironmental degradation, it also alters key behavioral,

psychological, and human factors of driving. These

factors include perceived safety, perceived stress, per-

ception of the driving environment, and driver con-

ﬁdence, all of which are often challenging to con-

ventional drivers. Therefore, the research questions

of the paper are as follows: 1) To what extent does

the introduction of autonomous shuttles in the study

area impact human factors such as perceived safety,

stress, perception, and driver conﬁdence? 2) What is

the quantitative relationship between perceived safety

and perceived stress, and how does perceived safety

inﬂuence driver stress levels?

This paper begins with a comprehensive literature

review, followed by a detailed methodology outlin-

ing the data collection, experimental design, simula-

tor setup, applied methods, and software used. The

results and discussion present key ﬁndings in relation

to the study’s objectives. Finally, the conclusion of-

fers an overall outlook and limitations of the study

and suggests potential areas for future research.

2 LITERATURE REVIEW

The following sections will introduce key topics rel-

evant to this study. The ﬁrst sub-section introduces

road safety and the leading causes of trafﬁc accidents

due to human errors. Further, the discussion will

delve into the inﬂuence of autonomous shuttles on

conventional driving behaviour and dynamics, with a

particular emphasis on how these shuttles affect car

driver’s perceptions of safety and stress. In the ﬁnal

section, the use of driving simulators to study these

behavioural shifts will be explored.

2.1 Road Safety and Accidents

The awareness of the trafﬁc environment is vital; hu-

man factors, such as general perception of the sur-

roundings, stress level, perception of safety, and con-

ﬁdence have a critical inﬂuence on a safer driving

environment. Research by Sayed et al. (2022) em-

phasized the inﬂuence of environmental conditions

and the importance of education and training to attain

awareness. Study ﬁndings showed that a challenging

trafﬁc environment induces high-stress levels and in-

creases driving errors. Further, points out that young

male drivers have a low perception of risk correlating

to accident rates.

Similarly, Xiao (2020) demonstrated that demo-

graphic factors, speciﬁcally age and gender, have a

crucial role in inﬂuencing accident rates, largely due

to differences in perception accuracy. Notably, Ta-

ble 1 depicts various human factors as a cause of ac-

cidents. 90% of road fatalities out of 1.19 million

incidents are seen in low and middle-income coun-

tries, primarily due to a lack of appropriate education,

training and awareness (World Health Organization,

2023a). Additionally, Lee and Winston (2016) under-

scored that drivers experience heightened stress levels

in situations where they feel a lack of control or low

perception of safety and also highlighted behavioural

implications, such as abrupt braking or quick lane

change in response to increased stress levels.

2.2 Autonomous Shuttles

Autonomous shuttles possess numerous advantages

in contrast to conventional buses, especially in

metropolitan areas. A study by Bucchiarone et al.

(2020) suggested that they could promote shared

and integrated transportation, improve tourism expe-

riences, and offer last-mile deliveries. In addition,

Giese and Klein (2005) illustrated that these shut-

tles are an integral part of Intelligent Transportation

Systems (ITS), which combines on-demand schedul-

ing, cost-effective techniques and efﬁcient usage of

resources. Intelligent autonomous shuttles are able to

transport a small amount of goods or a small group of

passengers on their own. They can also independently

navigate a passive track system and make decentral-

ized operational decisions (Giese and Klein, 2005).

Further, Bucchiarone et al. (2020) emphasized

how autonomous shuttles, speciﬁcally when deployed

in designated lanes, have the potential to increase road

safety, decrease emissions, and improve trafﬁc efﬁ-

ciency. The case study results highlighted the poten-

tial of autonomous shuttles to serve hard-to-reach ar-

eas such as city centers, corporate headquarters, and

A Simulator Study on Car User’s Perceptions in Interaction with Autonomous Shuttles

Table 1: Global Road Accident Data by Human Error Factors.

Human Error Fac-

tor

Key Statistics Source

Limited Awareness Globally, road trafﬁc fatalities reach approximately 1.19

million annually, with human error, mainly perception

and recognition failures, accounting for about 41% of in-

cidents.

(World Health Orga-

nization, 2023a)

Stress Level Limited perceived safety in environments with larger ve-

hicles, intersections, and close-following trafﬁc can in-

crease stress, elevating heart rates and anxiety, which are

linked to error-prone driving behaviors.

(Maga

na et al., 2020;

Tavakoli et al., 2023;

Lee and Winston,

2016)

Distracted Driving Distracted driving accounts for approximately 8-10% of

global road fatalities, with a high prevalence of mobile

phone use.

(World Health Orga-

nization, 2023a)

Speeding Nearly 30% of global trafﬁc deaths are attributed to

speeding; each 1% speed increase raises fatality risk by

4%.

(World Health Orga-

nization, 2023c)

Fatigued Driving Fatigue-related crashes cause signiﬁcant global fatalities;

fatigue equates to a 0.08% BAC impairment level.

(World Health Or-

ganization and Na-

tional Sleep Founda-

tion, 2023)

Impaired Driving

(Alcohol/Drugs)

Driving under inﬂuence increases crash fatality risk up to

5x; alcohol impairment is a factor in around 15-20% of

global fatalities.

(World Health Orga-

nization, 2023b)

hospitals, where traditional buses may face limita-

tions and the potential possibility that autonomous

shuttles will completely change urban transportation

(Bucchiarone et al., 2020). However, Iclodean et al.

(2020) pointed out the necessity to consider some

of the legal frameworks and social aspects following

the widespread use of autonomous driving as a pub-

lic transportation method associated with it (Iclodean

et al., 2020).

Nevertheless, there are still knowledge gaps about

the usage of such newer mobility options. The lack of

availability of a driver or an operator inside the vehi-

cle might raise trust-related questions and show a low

acceptance compared to conventional human-driven

vehicles. Research by Aramrattana et al. (2022)

found that when sharing the road with autonomous

shuttles, several notable adaptations have been ob-

served. First, drivers tend to adjust their following

distance, often reducing the time gap during on-ramp

scenarios but increasing it on highways when driv-

ing near autonomous vehicles (AVs). Furthermore,

lane-changing frequency decreases, suggesting that

conventional drivers become more conservative when

driving alongside AVs. Similarly, a reduction in over-

all speed has also been observed (Aramrattana et al.,

2022).

2.3 Simulator Studies

Driving simulators are crucial for many reasons, and

they have great signiﬁcance in research, training, test-

ing, and decision-making concerning trafﬁc actions.

Simulators help to build models closer to reality and

perform various experiments in a safe and controlled

manner. A key advantage of driving simulator stud-

ies lies in their cost-efﬁciency compared to real-world

vehicle testing. Physical vehicle testing often de-

mands substantial expenses, particularly as vehicles

may require tuning or modiﬁcations to suit each sce-

nario. Since research typically encompasses multi-

ple scenarios, incorporating various sets of test ve-

hicles for each is not cost-effective. Moreover, con-

ventional vehicles may not consistently perform as

intended within controlled conditions, adding fur-

ther complexity to accurate scenario-based testing.

Additionally, there is a possibility for unlimited it-

erations without damaging or disturbing infrastruc-

ture and trafﬁc ﬂow. Wang et al. (2007) reviewed

seven aspects in a study, which include driving be-

haviour studies, driver education and training, trans-

portation infrastructure, medicine and therapy, er-

gonomics, intelligent transportation system, and ad-

ministrative method and these aspects have been eval-

uated efﬁciently with the help of a driving simulator.

Additionally, in therapy and medicine, driving sim-

VEHITS 2025 - 11th International Conference on Vehicle Technology and Intelligent Transport Systems

ulators are used to evaluate and rehabilitate people

with driving-related disabilities (Wang et al., 2007).

Another study by Santiago-Chaparro et al. (2011) of-

fered valuable insights into road safety audits using

driving simulators. It proposes conceptual procedures

to speed up and enhance safety audits by introducing

programming and other software extensions that can

signiﬁcantly reduce time. However, this cannot be ap-

plied globally due to a factor called simulator sickness

(Santiago-Chaparro et al., 2011).

Simulator sickness is an essential consideration

in any simulator study. Symptoms caused by mo-

tion sickness include nausea, dizziness, headache and

disorientation, which impact the participant’s health.

Ethical and legal considerations are necessary, and a

comfortable experimental design must be modelled to

minimize Simulator sickness. Numerous researchers

refer to simulator sickness with different terms. For

instance, Brown et al. (2022) calls it cybersickness,

and Kennedy et al. (1993) terms it simulator sickness.

Likewise, this study will make use of the term simula-

tor sickness. When participants feel unwell or uncom-

fortable during the study, they might choose to leave,

posing challenges in keeping them engaged for the

duration of the research. Additionally, such unpleas-

ant experiences can reduce their enthusiasm and will-

ingness to participate, impact travel behaviour signif-

icantly and engagement levels within the study, and

raise concerns about data quality and reliability. The

cause of simulator sickness is still debated, and many

hypotheses and theories have been discussed.

2.4 Research Gap and Contribution of

this Study

Evaluating changes in driving dynamics and

analysing the variations with human factors are piv-

otal for the integration of autonomous shuttles, which

host many advantages over conventional options in

the existing ecosystem. However, it is necessary to

understand its impact and the fundamental changes

required to accommodate such mobility options.

Garus et al. (2022) supports the fact that new mo-

bility solutions are transforming road transportation,

such as autonomous vehicles and have the potential to

impact travel behaviour signiﬁcantly. The researcher

further points out that it is challenging to incorpo-

rate these changes into traditional models. The de-

gree to which behavioural changes are taken into ac-

count determines how well the existing models per-

form (Garus et al., 2022).

Additionally, the experimental design should be

optimized to eliminate any potential redundancy, en-

suring maximum efﬁciency and precision in the data

collection process. A study by Gold et al. (2018)

proposed a taxonomy for testing scenarios in human

factors research of Level 3 automated vehicles. Fur-

ther, a study by Calvi et al. (2022) highlighted how

spatial constraints, such as road layout, affect driv-

ing behaviour. For instance, in tighter spaces or ar-

eas with reduced lanes (e.g., from three lanes to one),

the driver’s behaviour shifts dramatically when fol-

lowing autonomous vehicles. Similarly, the study

emphasized that trafﬁc volume and spatial conditions

are critical in the assessment of behavioural changes

(Calvi et al., 2022)

In summary, this paper attempts to critically ex-

amine the impact of autonomous shuttles on conven-

tional car drivers, considering human factors and their

psychological dimension along with identifying rela-

tive behavioural changes. Further, the research seeks

to have a closer view of some of these critical voids by

focusing speciﬁcally on driver’s perceptions concern-

ing the new mobility option in the study area, as this

will seemingly provide new knowledge and empirical

facts to the ﬁeld of ITS.

3 METHODOLOGY

This section discusses the overall design of the exper-

iment. As mentioned earlier, the study aims to ex-

amine the impact of the introduction of autonomous

shuttles on the driving behaviours of conventional car

users in the study area. A static driving simulator and

a qualitative survey questionnaire have been incorpo-

rated to investigate changes in the psychological re-

sponses of conventional car drivers when verbally in-

formed about the presence of an autonomous shuttle

in one of the scenarios. The survey was designed to

comprehend traditional car driver’s perspectives and

readiness to drive with such mobility options. The

results obtained by the qualitative survey were statis-

tically analyzed to obtain cumulative and ﬁnal results

from all the test subjects.

3.1 Study Procedure

In this subsection, speciﬁc hypotheses of interest will

be presented along with the design of the experiment

to validate or discard the presented hypotheses.

The ﬁrst hypothesis (H1) states that the presence

of an autonomous shuttle in the study area environ-

ment will inﬂuence the physiological and behavioural

aspects of conventional car users. Similarly, the sec-

ond (H2) emphasises that perceived safety directly

contributes to the levels of perceived stress in drivers.

The design of the experiment integrated a static

A Simulator Study on Car User’s Perceptions in Interaction with Autonomous Shuttles

simulator into this study to provide an immersive driv-

ing experience as shown in Figure 1, portrays a vir-

tual environment of the study area that has been de-

signed and developed using OpenDrive (Association

for Standardization of Automation and Measuring

Systems (ASAM), 2024) data to build a transporta-

tion network and related infrastructure. This data was

gathered from an earlier survey conducted in 2015 by

a private surveyor named 3D Mapping Solutions.

Figure 1: Virtual Environment.

Similarly, trafﬁc data has been collected and in-

corporated into the experimental design. The study

area includes a camera-based trafﬁc data collection

system to collect and capture trafﬁc statistics. Cam-

eras have been installed in various locations on the

campus to capture the movements of all vehicles and

trafﬁc participants. These cameras are installed at ma-

jor intersections, entry-exit points, and other hot spots

on campus.

Further, Figure 2 depicts the simulation unit incor-

porated in the study and Figure 3 illustrates the differ-

ent software and tools employed in the design, such

as DYNA4; it provides a platform to handle, access,

modify and perform the experiment, further offers

model conﬁguration and scenario management to cre-

ate, import or export models and attributes of Vehicle

under Test (VuT) (Vector Informatik GmbH, 2024b).

Similarly, CANoe is a versatile and essential soft-

ware interface; it provides a thorough framework for

creating, evaluating, and optimizing electronic con-

trol units (ECUs) and the complex network connec-

tions seen in automobiles. CANoe handles mechani-

cal aspects of the simulator, which includes collision

counter detection (Vector Informatik GmbH, 2024a).

Likewise, Matlab (The MathWorks, Inc., 2024a) and

Simulink (The MathWorks, Inc., 2024b) are core parts

of the simulation setup that support CANoe opera-

tions in the background. The block models required

to control the vehicle performance and simulator con-

ﬁgurations were provided and handled with Simulink.

Another powerful tool, SUMO, is a microscopic sim-

ulation tool that has been incorporated with the inter-

face of DYNA4 to enable and control trafﬁc demand

in the virtual environment. Each of these applications

performs speciﬁc tasks, and its integration facilitates

a working simulator environment.

Figure 2: Simulator Setup - Hardware.

Considering the trafﬁc volumes, road categories

and trafﬁc regulations inside the campus, a route

has been selected from point A to Point B, which

includes different road categories (main street and

shared space) and trafﬁc conditions and is also an im-

portant commuting route inside the study area. Fig-

ure 4 depicts the route selected along with camera

locations. The red arrow mark indicates the starting

point for the VuT. The total length of the route is 1.6

kilometres, including sections of the main street (red

line) and shared path (blue line). As the name sug-

gests, shared space can be used by all possible classes,

such as cars, trucks, bikes, motorbikes, e-scooters and

pedestrians. Vulnerable users hold the highest priority

on such roads. It also includes some major intersec-

tions and a low-speed zone in front of Kindergarten.

The simulation experiment includes two scenar-

ios: one with a conventional shuttle and another

with the same shuttle presented to participants as au-

tonomous. Both scenarios are otherwise identical,

allowing comparison of responses based on the per-

ceived automation level. Each of these scenarios has

been designed to study the interaction between VuT

and the respective shuttle. This study considered six

categories of vehicles for other road users: cars, bicy-

cles, e-scooters, motorbikes, pedestrians, and a bus,

often termed a shuttle as far as the study is concerned.

3.2 Participants and Experimental Task

The preliminary stage essentially includes a brieﬁng

session of 15 minutes that lays out a general overview

of the experiment and essential tasks expected from

the participants. A total of 23 participants were in-

volved in the study: 10 female and 13 male candi-

VEHITS 2025 - 11th International Conference on Vehicle Technology and Intelligent Transport Systems

Figure 3: Simulator Setup - Data and Software Tool.

Figure 4: Route and Camera Location.

dates aged between 22 and 40 years with a mean driv-

ing experience of ten years (SD = 4.743). The pool of

participants had different demographic backgrounds,

consisting of nine nationalities, and everyone had at

least a bachelor’s degree.

Figure 5 narrates the procedures involved with the

experimental tasks. In the ﬁrst step, demographic de-

tails of the participants were acquired through a set of

comprehensive demographic questionnaires. These

surveys will be answered via an online tool, SociSur-

vey. The survey includes a declaration of consent,

information about data protection, information about

the simulator, and tasks to be performed.

Further, each participant will be trained in a spe-

ciﬁc training scenario to get acquainted with the dy-

namics of the simulator, followed by a simulator sick-

ness questionnaire (SSQ) designed by Kennedy et al.

(1993). The SSQ results are pivotal in assessing the

participant’s condition in regard to simulator sickness.

Once the test participants felt comfortable with the

simulator, they were introduced to the testing scenar-

ios. As Figure 5 depicts, participants were let to drive

in Scenario 1 along with a conventional shuttle in the

scenario. Similarly, in Scenario 2, before initiating

Scenario 2, a piece of new information was given to

the participants verbally ’This scenario is with the

A Simulator Study on Car User’s Perceptions in Interaction with Autonomous Shuttles

Figure 5: Flow Diagram - Experimental Phase.

Autonomous shuttle’. Additionally, participants an-

swered a qualitative survey after each scenario run.

The scenario questionnaire (qualitative) is a se-

ries of structured questions to collect the participant’s

opinions and perceptions about operating the driving

simulator in different scenarios. It had a particular

emphasis on interactions with autonomous shuttles.

Speciﬁc questions were designed to measure the per-

ceived safety, stress levels, and ability to predict the

behaviour of autonomous shuttles. In the end, SSQ

has been exercised again to examine the degree of

simulation sickness.

4 RESULTS AND DISCUSSIONS

This section presents the study outcome and data ﬁl-

tering procedures. The results should support or con-

tradict the initial hypothesis on how the introduction

of autonomous shuttles affects the behaviour of car

drivers and safety. Further, to enhance the under-

standing of the results concerning both psychologi-

cal and behavioural aspects focused in this study, the

outcomes have been analysed and addressed in two

sections.

4.1 Outcome Filtering

A collision counter has been incorporated as an in-

dicator with a pre-deﬁned reference limit; based on

the sample size of test subjects, a boundary limit is

assumed and deﬁned. The assumptions were made

considering historical results and an iterative testing

procedure.

The collision counter is a simulator-inclusive sig-

nal that detects the number of collisions between VuT

and the rest of the trafﬁc in each scenario. This data

is not hidden from participants. Any collisions be-

tween VuT and road users are instantly conveyed to

the driver via a vibration signal on the steering wheel.

The participants with more than two collisions per

scenario were excluded from the dataset. However,

only one participant was excluded from the results

dataset due to erratic driving and high collision en-

counters. The remaining 22 participants were seen

with zero or less than two collisions.

4.2 Statical Analysis of Qualitative

Results

Mann Whitney U test, also known as the Wilcoxon

rank-sum test, has been employed in the result anal-

ysis of this study. Mann-Whitney U test is a non-

parametric statistical test used to determine if there

are signiﬁcant differences between two independent

groups. It is particularly useful when comparing two

groups with ordinal data or continuous data with a

small sample size test that does not require normal

distribution, making it appropriate for the study. Fur-

ther, Equation 1 and 2 present the mathematical for-

mulation of the Mann-Whitney U test, used to com-

pare the two independent groups based on their rank

sums.

= n

+ 1)

− R

(1)

Where:

is the Mann-Whitney U statistic for the ﬁrst group,

is the number of observations in the ﬁrst group,

is the number of observations in the second group,

is the sum of the ranks for the observations in the

ﬁrst group.

Similarly, for the second group, you can compute:

= n

+ 1)

− R

(2)

VEHITS 2025 - 11th International Conference on Vehicle Technology and Intelligent Transport Systems

Where:

is the Mann-Whitney U statistic for the second

group,

is the sum of the ranks for the observations in the

second group.

As mentioned in the earlier section, this study em-

phasizes three key factors: perceived safety, perceived

stress, and perception of the driving environment in

regard to conventional car drivers in the study area.

These factors are discussed in detail in the following

subsections.

Perceived Safety

Safety corresponds to safety ratings opted by par-

ticipants in regard to interaction with the autonomous

or conventional shuttle, depending on the scenarios.

The participants conveyed their opinions using the

seven-star Likert scaling used in the scenario survey.

Seven corresponds to higher safety, and one is rela-

tively unsafe.

Figure 6 indicates that participants generally per-

ceive higher safety when interacting with conven-

tional shuttles compared to autonomous shuttles. The

mean value for perceived safety is higher for conven-

tional shuttles. Also, Table 4 indicates the descrip-

tive statistics of two scenarios across 22 participants

(N=22), suggesting the conventional car drivers felt

comparatively safer (Mean: M = 5.909, Standard De-

viation: SD = 1.151) while driving along with the

conventional shuttle rather than an autonomous shut-

tle (M = 2.818, SD = 1.816). Even though the driving

behaviour of the autonomous shuttle in the simula-

tion environment is identical to the driving behaviour

of the conventional shuttle. The only difference be-

tween the two scenarios is that the participants are in-

formed beforehand that the shuttle is autonomous or

conventional.

Figure 6: Perceived Safety of Conventional Car Drivers.

Table 2, displays U-statistics; It represents the sum

of ranks for one of the two groups compared in the test

and p-statistics; A p-value less than 0.001 indicates

strong evidence against the null hypothesis, suggest-

ing a statistically signiﬁcant difference in perceived

safety between the two groups. Further, the Mann-

Whitney U test shows that the perceived safety is sig-

niﬁcant (p = 0.001< 0.05). Therefore, this supports

the ﬁrst hypothesis (H1) of the study.

Table 2: Independent Samples T-Test for Perceived Safety.

U p

Perceived Safety 49.000 < .001

Perceived Stress

Another indicator used in this study is stress,

which helps assess diverse study components. Per-

ceived stress serves as a multi-dimensional unit

inﬂuencing distinct aspects of transportation sys-

tems. Firstly, stress directly impacts driver behaviour.

Heightened stress levels can impair decision-making,

reaction times, and sustained attention, potentially

compromising driving safety.

As shown in Figure 7, stress perceived by con-

ventional users is similar in both scenarios. Further,

the stress scores indicated by descriptive details for

the scenario with the autonomous (M = 2.818, SD =

1.816) and conventional shuttles (M = 2.682, SD =

1.615) are relatively similar.

Figure 7: Perceived Stress of Conventional Car Drivers.

Additionally, in the Mann-Whitney U test, as

shown in Table 3, the perceived stress does not dif-

fer signiﬁcantly (p = 0.981> 0.05) and also, perceived

stress seemed not to inﬂuence the perceived safety to

the same extent. Therefore, the hypothesis (H2) is

rejected based on the previously chosen conﬁdence

level.

Table 3: Independent Samples T-Test for Perceived Stress.

U p

Perceived Stress 243.500 0.981

A Simulator Study on Car User’s Perceptions in Interaction with Autonomous Shuttles

Table 4: Grouped Descriptive Values for Human Factors.

Group N Mean SD SE Coefﬁcient

of variation

Perceived Safety

Autonomous

Shuttle

22 2.818 1.816 0.387 0.644

Conventional

Shuttle

22 5.909 1.151 0.245 0.195

Perceived Stress

Autonomous

Shuttle

22 2.818 1.816 0.387 0.644

Conventional

Shuttle

22 2.682 1.615 0.344 0.602

Perception

Autonomous

Shuttle

22 5.364 1.529 0.326 0.285

Conventional

Shuttle

22 5.727 1.518 0.324 0.265

Conﬁdence of Driving with AVs

Male 13 4.462 0.519 0.144 0.116

Female 9 3.556 0.882 0.294 0.248

Even though the factor of stress perception is stat-

ically insigniﬁcant, overall mean values (see Table 4)

suggest a slight change in the behaviours of the car

drivers as they tend to perceive rather in a different

manner while interacting with the autonomous shut-

tle.

Perception

Perception corresponds to awareness of the sur-

rounding trafﬁc environment, which is a vital indica-

tor of the psychological and behavioural aspects of

this study. The acceptance of new mobility options

and, therefore, the ability to understand the dynam-

ics of these mobilities become a meaningful part of

the campus transportation ecosystem is directly tied to

user acceptance. When individuals perceive such ve-

hicles and their operations as safe and reliable, adop-

tion increases, and the integration into the existing

transportation system proceeds smoothly.

As Figure 8 depicts, perceptions of both shuttle

options, autonomous and conventional shuttles, are

relatively similar. However, as shown by descriptive

statistics in Table 4, the autonomous shuttle tends to

be perceived slightly less (M = 5.364, SD = 1.529) by

the car drivers as compared to a conventional shuttle

(M = 5.727, SD = 1.518) within the exact same envi-

ronment.

Furthermore, as shown by Table 5, the Mann-

Whitney U test indicates that perception does not dif-

fer and is insigniﬁcant (p = 0.343> 0.05) with the

chosen conﬁdence levels. Therefore, perception as

a factor leads to the rejection of the ﬁrst hypothesis

(H1) in the study.

Additionally, the survey results highlight the con-

ﬁdence levels of drivers when sharing the road with

Figure 8: Perception of Conventional Car Drivers.

Table 5: Independent Samples T-Test for Perception.

U p

Perception 202.500 0.343

AVs in an urban environment. The demographic

data of the participants provided valuable insights into

these perceptions. After the data ﬁltering, the result

dataset contained 22 participants, consisting of nine

female and 13 male drivers, who shared their views

on driving alongside AVs.

Figure 9 illustrates that male drivers reported

higher conﬁdence when interacting with AVs in an ur-

ban environment, with an average score of M = 4.462

(SD = 0.592). In contrast, female drivers expressed

lower conﬁdence, resulting in a comparatively lower

average score of M = 3.556 (SD = 0.294) under the

same trafﬁc conditions.

Further, Table 6 indicates the value from the

Mann-Whitney U test. The difference between the

VEHITS 2025 - 11th International Conference on Vehicle Technology and Intelligent Transport Systems

Figure 9: Conﬁdence Levels of Conventional Car Drivers.

Table 6: Independent Samples T-Test for Conﬁdence of

Driver Interacting with AVs.

U p

Conﬁdence of Driving

with AVs

93.000 0.013

two groups is statically insigniﬁcant (p = 0.013>

0.05) given the conﬁdence interval of 95%. Never-

theless, female participants expressed varied conﬁ-

dence across the rating scale, wherein the male drivers

tended to report relatively higher conﬁdence.

The ﬁndings of this research demonstrate, that

perceived safety emerged as a key metric, reﬂect-

ing the psychological and behavioral impacts of au-

tonomous shuttles on conventional car drivers. The

distinct differences in safety perceptions between au-

tonomous and conventional shuttles highlight its cen-

tral role in shaping driver interactions within mixed-

trafﬁc environments. While perceived safety was a

primary focus, other interconnected factors, such as

stress levels, driver conﬁdence, and the perception of

the driving environment, provided essential insights

that enriched the analysis. Collectively, these aspects

form a comprehensive framework for understanding

how emerging autonomous technologies affect human

factors, offering a foundation for a scalable frame-

work for assessing human factors in similar experi-

mental setups.

5 CONCLUSIONS AND FUTURE

WORK

Overall, the introduction of autonomous shuttles is

discussed, along with the crucial notion of integrating

emerging mobility solutions into the driving environ-

ment of the study area as an essential component of

the reshaping of mobility.

The study effectively demonstrates the inﬂuence

of autonomous shuttles on driving behaviours, per-

ceptions, and driver psychology. Findings revealed a

signiﬁcant reduction in perceived safety levels when

drivers were informed verbally about the autonomous

shuttle’s presence, even though both scenarios were

otherwise identical to each other. It also highlights

that the perceived safety and perceived stress of a con-

ventional car driver are not directly related to interact-

ing with autonomous shuttles within the deﬁned traf-

ﬁc environment.

Similarly, the variations across different human

factors with different participants throughout the

study clearly point out that human perceptions sig-

niﬁcantly impact driving, and deviations indicate the

degree of perception change in reference to aggre-

gated mean driving behaviours. Even though the re-

sult cannot be generalized to the whole driving com-

munity. The study, which was conducted effectively

with 22 participants, resulted in slightly varied per-

ceptions and lesser safety perception levels concern-

ing autonomous shuttles. Nevertheless, no large-scale

or fundamental issues have resulted in the introduc-

tion of autonomous shuttles on campus.

Though the study still has certain limitations,

these can be tackled in future research designs with

appropriate considerations. A much bigger sample

size with varied demographic backgrounds might help

in a deeper analysis of driving perceptions. Simi-

larly, consideration of other human and psychological

factors might provide a closer understanding of be-

havioural data. Detailed geometric data covering all

aspects of road infrastructure on the campus will en-

hance visualization of the interaction between trafﬁc

and VuT. Further research direction includes the inte-

gration of vehicles-to-everything (V2X) communica-

tions units and connected simulators.

In conclusion, along with the results, the study of-

fers a framework to design, develop and perform an

experiment in regard to driving behaviours and sur-

vey procedures. The complete experiment is designed

from scratch to study the impact of autonomous shut-

tles on the campus. The same experimental design

can be adapted for further research with other vehi-

cle types as well, considering appropriate and cali-

brated driving models. Additionally, the procedure

discussed to carry out an experiment with participants

considering various aspects, the study had established

a comprehensive experiment design.

ACKNOWLEDGEMENTS

The research was carried out under the Munich

Mobility Research Campus (MORE) project with

A Simulator Study on Car User’s Perceptions in Interaction with Autonomous Shuttles

the ﬁnancial support of the Deutsche Zentrum f

Digitalisierungs- und Technologieforschung der Bun-

deswehr (dtec.bw) under the funding of Europ

aischen

Union – NextGenerationEU. We would like to express

our gratitude to Prof. Dr. Silja Hofmann for their sup-

port and insights throughout the experiment. AI tool

Chat GPT has been used to improve the legibility of

the text.

REFERENCES

Aramrattana, M., Fu, J., and Selpi (2022). Behavioral adap-

tation of drivers when driving among automated vehi-

cles. Journal of Intelligent and Connected Vehicles,

5(3):309–315.

Association for Standardization of Automation and Mea-

suring Systems (ASAM) (2024). ASAM Open-

DRIVE Standard. Available at: https://www.asam.net/

standards/detail/opendrive/.

Bartneck, C., L

utge, C., Wagner, A., and Welsh, S. (2021).

Autonomous Vehicles, page 83–92. Dual Use Sci-

ence and Technology, Ethics and Weapons of Mass

Destruction.

Brown, P., Spronck, P., and Powell, W. (2022). The simu-

lator sickness questionnaire, and the erroneous zero

baseline assumption. Frontiers in Virtual Reality,

3:118.

Bucchiarone, A., Battisti, S., Marconi, A., Maldacea, R.,

and Ponce, D. C. (2020). Autonomous shuttle-as-a-

service (asaas): Challenges, opportunities, and social

implications. IEEE Transactions on Intelligent Trans-

portation Systems, 22(6):3790–3799.

Calvi, A., D’Amico, F., Ferrante, C., and Calcaterra, G.

(2022). A driving simulator study on the effects of

autonomous vehicles on drivers behaviour under car-

following conditions. Human Factors in Transporta-

tion, 60(60):10–54941.

Garus, A., Alonso, B., Alonso Raposo, M., Ciuffo, B., et al.

(2022). Impact of new mobility solutions on travel be-

haviour and its incorporation into travel demand mod-

els. Journal of Advanced Transportation, 2022.

Giese, H. and Klein, F. (2005). Autonomous shuttle system

case study. In Scenarios: Models, Transformations

and Tools: International Workshop, Dagstuhl Castle,

Germany, September 7-12, 2003, Revised Selected Pa-

pers, pages 90–94. Springer.

Gold, C., Naujoks, F., Radlmayr, J., Bellem, H., and

Jarosch, O. (2018). Testing scenarios for human fac-

tors research in level 3 automated vehicles. In Ad-

vances in Human Aspects of Transportation: Proceed-

ings of the AHFE 2017 International Conference on

Human Factors in Transportation, July 17- 21, 2017,

The Westin Bonaventure Hotel, Los Angeles, Califor-

nia, USA 8, pages 551–559. Springer.

Iclodean, C., Cordos, N., and Varga, B. O. (2020). Au-

tonomous shuttle bus for public transportation: A re-

view. Energies, 13(11):2917.

Kennedy, R. S., Lane, N. E., Berbaum, K. S., and Lilien-

thal, M. G. (1993). Simulator sickness questionnaire:

An enhanced method for quantifying simulator sick-

ness. The international journal of aviation psychol-

ogy, 3(3):203–220.

Lee, Y.-C. and Winston, F. (2016). Emotions & driving

in stressful trafﬁc situations. Transportation Research

Part F: Psychology and Behavior, 42:215–226.

Maga

na, V. C., Scherz, W. D., Seepold, R., Madrid, N. M.,

neda, X. G., and Garcia, R. (2020). The effects of

the driver’s mental state and passenger compartment

conditions on driving performance and driving stress.

Sensors, 20(18):5274.

National Highway Trafﬁc Safety Administration (2017).

Quick facts 2016. Technical Report DOT HS

812 456, U.S. Department of Transportation, Na-

tional Highway Trafﬁc Safety Administration. Avail-

able at: https://crashstats.nhtsa.dot.gov/Api/Public/

Publication/812456.

Santiago-Chaparro, K., DeAmico, M., Bill, A., Chitturi, M.,

and Noyce, D. (2011). Virtual road safety audits using

driving simulators: A framework. In 3rd International

Conference on Road Safety and Simulation.

Sayed, I., Abdelgawad, H., and Said, D. (2022). Study-

ing driving behavior and risk perception: a road safety

perspective in egypt. Journal of Engineering and Ap-

plied Science, 69(1):22.

Tavakoli, A., Lai, N., Balali, V., and Heydarian, A. (2023).

How are drivers’ stress levels and emotions associated

with the driving context? a naturalistic study. Journal

of Transport & Health, 31:101649.

The MathWorks, Inc. (2024a). MATLAB Documenta-

tion. Available at: https://www.mathworks.com/help/

matlab/.

The MathWorks, Inc. (2024b). Simulink Documenta-

tion. Available at: https://www.mathworks.com/help/

simulink/.

Vector Informatik GmbH (2024a). CANoe4SW:

Software-in-the-Loop Testing Platform. Avail-

able at: https://www.vector.com/int/en/products/

products-a-z/software/canoe4sw/.

Vector Informatik GmbH (2024b). DYNA4: Simula-

tion and Virtual Testing Platform. Available at:

https://www.vector.com/int/en/products/products-a-z/

software/dyna4/.

Wang, Y., Zhang, W., Wu, S., and Guo, Y. (2007). Simula-

tors for driving safety study–a literature review. In Vir-

tual Reality: Second International Conference, ICVR

2007, Held as part of HCI International 2007, Bei-

jing, China, July 22-27, 2007. Proceedings 2, pages

584–593. Springer.

World Health Organization (2023a). Global status report on

road safety 2023. Available at: https://www.who.int/

publications/i/item/9789240086517.

World Health Organization (2023b). Global sta-

tus report on road safety 2023. Avail-

able at: https://www.who.int/teams/

social-determinants-of-health/safety-and-mobility/

global-status-report-on-road-safety-2023.

VEHITS 2025 - 11th International Conference on Vehicle Technology and Intelligent Transport Systems

World Health Organization (2023c). Speed management

report 2023. Available at: https://www.who.int/

news-room/fact-sheets/detail/road-trafﬁc-injuries.

World Health Organization and National Sleep Foun-

dation (2023). Fatigue and road safety: A

global analysis. Global Studies on Fatigue in

Road Safety. Available at: https://www.thensf.org/

drowsy-driving-prevention/.

Xiao, Y. (2020). Analysis of the inﬂuencing factors of the

unsafe driving behaviors of online car-hailing drivers

in china. Plos one, 15(4):e0231175.

A Simulator Study on Car User’s Perceptions in Interaction with Autonomous Shuttles