AHP-based Evaluation of the Acceptance of Autonomous Driving

André Jacob

, Andreas Ahrens

, Jelena Zascerinska

and Cesar Benavente-Peces

Hochschule Wismar, University of Technology, Business and Design, Wismar, Germany

Centre for Education and Innovation Research, Riga, Latvia

Escuela Tecnica Superior de Ingenieria y Sistemas de Telecomunicacion,

Universidad Politecnica de Madrid, Crtra de Valenica, km 7, Madrid, Spain

Keywords: Autonomous Driving, AHP-based Evaluation, Utility Analysis, Extrinsic Factors, Intrinsic Factors, Other

Factors.

Abstract: In the world with seven billion people and around one billion cars, autonomous driving has the potential to

become a key factor of overcoming pollution and traffic jam when considering big cities. This work discusses

safety- and security related factors with simultaneous improving of the acceptance of autonomous driving.

The data were obtained from the survey carried out by Deloitte analytics institute located in Germany. The

data are analysed by means of the analytical hierarchy process (AHP). The theoretical finding allows

concluding that, due to the technology development, the conventional division of factors into external and

internal has to be extended with the third groups, namely other factors. A theoretical finding is three clusters

of influencing factors. The cluster with the most influential factors includes active driving systems, sensors,

car-to-car communications and algorithmic. The limitations of the study are outlined. Directions of further

work are proposed.

1 INTRODUCTION

Autonomous driving is one of the greatest goals in the

automotive industry. It can be considered as an

innovative component of the future of car driving as

the whole car industry. Autonomous driving has

become a key research in the last decade. There are

several manufacturers, which do investigations in the

field of autonomous driving in order to create an

innovative autonomous driving car for their

customers.

In the world with seven billion people and around

one billion cars, autonomous driving has the potential

to become a key factor of overcoming pollution and

traffic jam when considering big cities. Here,

autonomous cars using hydrogen fuel for motive

power or even electric cars can have the potential to

improve the living conditions in big cities

significantly.

It is assumed that autonomous cars can drive

much faster and at the same time more economically

in a convoy while avoiding annoying downtimes.

This should also drastically reduce fuel consumption.

On the other hand, it might also decrease the stress

level for the passengers as autonomous vehicles

"know" where the next traffic jam is and drive around

it independently or they know already in advance

where the next free parking space is available in

crowded city centres. An autonomous car would

search for a parking space on its own.

However, many people still have doubts

regarding the future of autonomous driving.

In previous work some authors examined

different aspects of acceptance of autonomous

driving. For example, Lee, Chang and Park (2018)

point out that

the most important influencing factors

for general autonomous vehicles South-Korean

costumers characterize as usefulness, reliability and

legality. The acceptance of full autonomous vehicles

is mostly affected by safety, user convenience, and

extra expenses. Safety also is important for the

acceptance of partial autonomous vehicles (Lee,

Chang, Park, 2018). In addition, Kettles and van Belle

(2019) show in their work, that Performance

Expectancy and Hedonic Motivation (as well known

as Enjoyment) are the most significant predictors of

Behaviour Intention.

The study of the acceptance of autonomous

driving is related to the examination of the supporting

trust of autonomous driving. Häuslschmid, von

Jacob, A., Ahrens, A., Zascerinska, J. and Benavente-Peces, C.

AHP-based Evaluation of the Acceptance of Autonomous Driving.

DOI: 10.5220/0010144100610065

In Proceedings of the 10th International Conference on Pervasive and Parallel Computing, Communication and Sensors (PECCS 2020), pages 61-65

ISBN: 978-989-758-477-0

Bülow, Pfleging and Butz (2017) carried out a

research on this topic. They argue, that trust in

autonomous driving could be increased by a

visualization, which represent the car’s interpretation

of the current situation and its corresponding actions

(Häuslschmid, von Bülow, Pfleging, Butz, 2017).

Moreover, customers have a conflicting attitude

about self-driving vehicles. The further development

of this technology is connected to some kind of

“posthuman ability” as well as it is connected to

hesitation to cede control to machines. Gambino and

Sundar (2019) examined exactly this topic. They

suggest, that “individuals are much more accepting of

technology that can clearly outclass human abilities”

(Gambino, Sundar, 2019).

Against this background, there are some aspects

such as safety- and security related questions that

have not received a proper attention in the

implemented research. This work discusses safety-

and security related questions with simultaneous

improving the acceptance of autonomous driving.

The data obtained from surveys are analysed by

means of the analytical hierarchy process (AHP).

This procedure was introduced by Saaty in the 70’s to

analyse complex decisions (Saaty, 2008).

The paper is structured as follows: Section 2

introduces the most important factors of autonomous

driving classified into intrinsic and extrinsic factors

as well as the other factors. Within these factors it will

be examined, how the influencing factors affect

security- and safety related questions in the

acceptance of autonomous driving. In Section 3 AHP

will be shortly introduced followed by a subsequent

weighting of the defined intrinsic and extrinsic factors

with respect to safety and security of autonomous

driving. Section 4 is dedicated to the analysis of the

acceptance of autonomous driving in our world today

using AHP. Finally, some concluding remarks are

given.

2 FACTORS INFLUENCING

AUTONOMOUS DRIVING

In this section, factors improving the acceptance of

autonomous driving are specified. By factor, a reason

of phenomenon change is meant. In the present work,

it is assumed that factors such as energy consumption,

the possibility to use active driving systems and Car-

to-Car as well as Car-to-infrastructure play an

important role when improving the acceptance of

autonomous driving.

Factors are traditionally differentiated into

external and internal (Ahrens, Zaščerinska, 2014). In

the present research, the factors’ influence is

classified in intrinsic, extrinsic and the other factors

as shown in Figure 1.

It should be pointed that in the present work the

“external” and “extrinsic” as well as “internal” and

“intrinsic” factors are used synonymously. Whereas

intrinsic factors are assigned to the inner area of the

vehicle such as the possibility of using active driving

systems, extrinsic parameters are assigned with the

outer area of the vehicle such as the energy

consumption.

Figure1: Factor classification.

Factors which cannot be assigned to intrinsic and

extrinsic are put into the class “the other factors”. In

this category will put factors influencing the Car-to-

Car as well as Car-to-infrastructure communication.

2.1 Intrinsic Factors

Intrinsic factors are assigned to the inner area of the

vehicle. The intrinsic factors include

 the behaviour of the vehicles (algorithmic

factors),

 physical components of driving, for example

breaks and the engine,

 the communication between the vehicle and a

mobile device like smartphone or tablet.

These factors are in particular physically located,

or work to a large extent in the inner area of the

vehicle.

It is not possible to create a safe autonomous

driving system by ignoring this kind of influencing

factors.

2.2 Extrinsic Factors

Extrinsic factors are assigned to the outer area of the

vehicle.

In this regard the extrinsic factors are

 cheap energy consumption (i.e. prices for gas,

electricity or hydrogen),

PECCS 2020 - 10th International Conference on Pervasive and Parallel Computing, Communication and Sensors

 sensors (possibility to perceive the vehicle’s

surroundings, for example light detection and

ranging (LIDAR, Sonar or radar),

 locating possibilities (techniques to define the

position of the vehicle exactly, for example

with the help of GPS and/or Galileo) and

 service infrastructure (the availability of gas

stations and car wash stations etc.).

These influencing factors are in particular located

in the outer area of the vehicle. As highlighted in

section 2.1 when analysing intrinsic factors, it is not

possible to create a safe autonomous driving system

by ignoring the extrinsic influencing factors.

2.3 The Other Factors

Other factors cannot be assigned to the classes of

intrinsic and extrinsic factors. This category includes

 Car-to-Car communication (standardized

communication between vehicles),

 Car-to-Infrastructure communication

(standardized communication between

vehicles and infrastructure like traffic lights)

and

 Car-to-Remote site (possibility to

communicate with the workplace or home)

3 UTILITY ANALYIS AND

ANALYTIC HIERARCHY

PROCESS

After considering the factors of security- and safety

issues related to autonomous driving it is essential to

analyse today’s acceptance of autonomous driving

and the potential of the influencing factors to improve

the acceptance of autonomous driving. There are

several ways to examine the aspect of autonomous

driving, especially the acceptance of autonomous

driving vehicles. A suitable tool to identify this

potential is a survey-based utility analysis.

Additionally, an analytic hierarchy process is used to

prioritize the influencing factors.

The utility analysis is divided into the phases:

firstly, to identify the main criteria of autonomous

driving, then, to prioritize these criteria and, finally,

to link them to the influencing factors.

The main criteria of autonomous driving are

identified as Conformity gain, Safety, Security,

Degree of automation increase and Costs. In order to

determine the importance of these criteria the

probands had to prioritize them against each other.

Percentage describes the importance of these criteria.

The results are:

 Conformity gain – 15%,

 Safety – 35%,

 Security – 35%,

 Degree of automation increase – 5%,

 Costs – 10%.

In addition to that, the probands judged the impact

of the influencing factors to these criteria. The results

of the utility analysis are shown in Table 1.

The average potential percentage is 64,25%. The

influencing factors which reach a higher percentage

are algorithmic, active driving systems, sensors,

locating, communication with mobile devices, car-to-

car communication and system integration in the

vehicle. The highest utility reach active driving

systems and sensors with 78% each.

Table 1: Potential of the influencing factors to improve the

acceptance of autonomous driving.

Influencing factor Potential [%]

Algorithmic 73

Car-in-the-Cloud 64

Cheap Energy 39

Active driving systems 78

Sensors 78

Locating 67

Communication with mobile

devices

Car-to-Car communication 68

Car-to-Infrastructure

communication

Car-to-Remote site 61

System integration in the vehicle 67

Service infrastructure 54

Table 2: Priority percentage of the influencing factors.

Influencing factor Potential [%]

Algorithmic 16,03

Car-in-the-Cloud 2,26

Cheap Energy 1,65

Active driving systems 21,05

Sensors 9,87

Locating 6,01

Communication with mobile

devices

2,87

Car-to-Car communication 12,92

Car-to-Infrastructure

communication

13,51

Car-to-Remote site 3,00

System integration in the vehicle 9,07

Service infrastructure 1,75

Analytic hierarchy process is a tool for a

reasonable decision making. The working method is

aimed at creating a matrix in which the rows and

AHP-based Evaluation of the Acceptance of Autonomous Driving

columns are the influencing factors. After that it is

necessary to compare every pair of influencing

factors. Determining an (to one standardized)

eigenvector is the step to generate the factors priority

percentage (Saaty, 2008). The results of the analytic

hierarchy process are shown in the Table 2.

4 IMPROVEMENT OF

ACCEPTANCE OF

AUTONOMOUS DRIVING

The survey shows that the survey participants are

interested in autonomous driving as 85% of the

respondents state that autonomous driving (full

automation) is going to be pushed through in the

future (Deloitte analytics institute, 2016).

In addition, three out of four groups of the

participants point that they are interested in testing

new innovations in the field of autonomous driving.

(Deloitte analytics institute, 2016). Approximately

50% of the passengers would like to do other things

while driving, for example checking mails, talk to

other passengers or sleep (Deloitte analytics institute,

2016).

The influencing factors of the high percentage are

active driving systems and algorithmic. Also, Car-to-

Car communication and Car-to-Infrastructure

communication are above 10% (see Table 2). The

highest potential reaches active driving systems with

21,05% (see Table 2). The consistency ratio of the

analytic hierarchy process is 0,074. Therefore, the

analytic hierarchy process can be seen as consistent,

and its results can be considered as usable.

However, all the described groups of respondents

would not be fine with a completely autonomous

driving vehicle (Deloitte analytics institute, 2016).

Therefore, the acceptance of autonomous driving

could be improved. A number of actions for the

improvement of the acceptance of autonomous

driving in combination with security and safety-

related questions could be elaborated. For the

determination of actions to improve the acceptance of

autonomous driving, the influencing factors are

clustered into three groups. These groups represent

the prioritization of the influencing factors possibility

to improve the acceptance of autonomous driving.

First of all, a rating system has to be implemented.

For each study (security-safety study, utility analysis

and analytic hierarchy process) every influencing

factor got a rating between 12 and one, describing

how important it is in this specific study. After that,

the sum of these three ratings is formed, and the

clusters are created. One cluster is created for each of

the four high-rated, middle-rated and low-rated

influencing factors. Table 3 shows the results of

clustering. Cluster 1 receives 121 rating-points,

which describes exactly 50% of the possible influence

on autonomous driving.

Table 3: Cluster of influencing factors.

Influencing factor Cluster

Algorithmic

(121 rating-

points)

Car-to-Car communication

Sensors

Active driving systems

Car-to-Infrastructure

communication

(80 rating-points)

Locating

Communication with mobile

devices

System integration in the

vehicle

Cheap Energy

(41 rating-points)

Car-to-Remote site

Car-in-the-Cloud

Service infrastructure

In addition, there are some synergies between

Cluster 1 and Cluster 2. For example, car-to-car

communication and car-to-infrastructure

communication using the same technical standards.

This shows, that it is possible to influence even more

than 50% by improving the factors of Cluster 1.

Actions, whose goal is to improve the acceptance

of autonomous driving in combination with security

and safety-related issues, should work in Cluster 1.

Specifically, they should affect one or more of the

influencing factors such as algorithmic, car-to-car

communication, sensors or active driving systems.

5 CONCLUSIONS

The theoretical finding of the research allows

concluding that, due to technology development, the

conventional division of factors into external and

internal has to be extended with the third groups,

namely the other factors. Another theoretical finding

is the clusters of influencing factors.

The empirical finding is the cluster with four most

influencing factors or, in other words, Cluster 1. This

cluster includes active driving systems, sensors, car-

to-car communications and algorithmic. These four

influencing factors combine 121 rating-points (50%

of all rating-points). These four factors obtain the

higher impact. Another finding reveals synergies

between the clusters. The empirical finding is that the

PECCS 2020 - 10th International Conference on Pervasive and Parallel Computing, Communication and Sensors

group of the other factors has the highest impact on

security- and safety related issues. This finding

proposes that the most important security- and safety

related problems will appear in one of the influencing

factors of this group. Another conclusion could be

drawn about influencing factors with a higher impact:

the influencing factors with a higher impact should be

firstly considered for the improvement of the

acceptance of autonomous driving.

The present work has some limitations. The

limitation is the use of the data of only one survey.

Another limitation is that the survey was carried out

only in one country, namely Germany.

Research on autonomous driving is attracting a

lot of interest in the scientific community. However,

autonomous driving is definitely under-investigated

and not sufficiently presented. There is still a lot to

investigate and discuss in the field of the acceptance

of autonomous driving and autonomous driving itself.

Consequently, the acceptance of autonomous driving

and autonomous driving are inter-related. The authors

draw the conclusion that the increase in the

acceptance of autonomous driving will promote the

development of autonomous driving on the whole.

Increased research efforts in the field of research on

the acceptance of autonomous driving will assist in

developing autonomous driving from a number of

aspects and perspectives.

Future work will investigate and compare the

relation between the present research and similar

works in the scientific literature. Adoption of

different technologies for autonomous driving will be

analyzed in further work.

Other typical analytical approaches will be

compared to the AHP algorithm. Further work will

also be devoted to the description of the calculation

process of the AHP algorithm. The search for other

approaches and methods to investigate the acceptance

of autonomous driving is proposed.

The further research tends to re-examine factors

that influence the acceptance of autonomous driving

as along with the technology development, new

factors could emerge.

Future research will also focus on the description

and analysis of case studies that can help further

elaborate the analytic process detail.

Future work also implies the utilization of proper

techniques for data collection in order to obtain a

relevant description of the contemporary situation of

the acceptance of autonomous driving. In these terms,

the focus could be more put on the application of

qualitative methods for a deeper analysis of

influencing factors.

Another research direction is to involve more

respondents into the study of the acceptance of

autonomous driving.

Insights about how the acceptance is speeding or

not in comparison with other technologies, in light of

current progress and events will be formulated in

future work.

A comparative study on the acceptance of

autonomous driving of different countries could be

interesting for the research community as well.

REFERENCES

Ahrens, A., Zaščerinska, J. (2014). Factors that Influence

Sample Size in Educational Research. ATEE Spring

University proceedings Changing Education in a

Changing Society, pp. 19-32. Klaipeda University.

Deloitte analytics institute. (2016). Autonomes Fahren –

wie Kunden in Deutschland überzeugt werden.

Retrieved from https://www2.deloitte.com/content/

dam/Deloitte/de/Documents/consumer-industrial-prod

ucts/Autonomes-Fahren-komplett-safe-Sep2016.pdf

Gambino, A., Sundar, S.S. (2019). Acceptance of Self-

Driving Cars: Does Their Posthuman Ability Make

Them More Eerie or More Desirable?. Glasgow.

Extended Abstracts of the 2019 CHI Conference on

Human Factors in Computing Systems

Häuslschmid, R., von Bülow, M., Pfleging, B., Butz, A.

(2017). Supporting Trust in Autonomous Driving.

Limassol. IUI '17: Proceedings of the 22nd

International Conference on Intelligent User

InterfacesMarch 2017 Pages 319–329 https://doi.org/

10.1145/3025171.3025198.

Kettles, N., van Belle, J.-P. (2019). "Investigation into the

Antecedents of Autonomous Car Acceptance using an

Enhanced UTAUT Model," 2019 International

Conference on Advances in Big Data, Computing and

Data Communication Systems (icABCD), Winterton,

South Africa, 2019, pp. 1-6, doi: 10.1109/

ICABCD.2019.8851011.

Lee, J., Chang, H., Park, Y.I. (2018). "Influencing Factors

on Social Acceptance of Autonomous Vehicles and

Policy Implications," 2018 Portland International

Conference on Management of Engineering and

Technology (PICMET), Honolulu, HI, 2018, pp. 1-6,

doi: 10.23919/PICMET.2018.8481760.

Rauch, S., Aeberhard, M., Ardelt, M., & Kämpchen, N.

(2012). Autonomes Fahren auf der Autobahn – eine

Potentialstudie für zukünftige Fahrerassistenzsysteme.

Corpus ID: 188629588. Retrieved from

https://www.semanticscholar.org/paper/Autonomes-

Fahren-auf-der-Autobahn-–-eine-für-Rauch-

Aeberhard/6342cbc16ee2bdfcaefc40ddcbdbc8daf61c2

070 .

Saaty, T.-L. (2008). Decision making with the analytic

hierarchy process. Pittsburgh: Int. J. Services Sciences,

Vol. 1, No. 1.

AHP-based Evaluation of the Acceptance of Autonomous Driving