11
1. INTRODUCTION
Personal mobility is in the midst of an unprecedented revolution lead by technological
transformations and improvements in the automotive industry. Autonomous driving
represents the future of transportation and its applications will have an impact not only on the
automotive sector but also on many aspects of society.
The common idea is that self-driving vehicles are a recent phenomenon, an innovation of the
21
st
century. However, autonomous vehicles are rooted in the 1939 Futurama exhibit
designed by General Motors at the New York World’s Fair, which showed their vision of an
automated highway system that would have guided self-driving cars. From that day on, many
steps forward have been made. Many organizations, ranging from car manufacturers to
startups founded for self-driving technologies, are working on the complete automation of the
automotive sector. However, self-driving vehicles represent a challenge for many automakers,
which will result in few players capable of building this technology and in a large
consolidation, as more and more car companies are developing partnerships with autonomous
driving startups. Examples of these partnerships include Waymo and Jaguar, and General
Motors and Cruise
This thesis is the result of careful and constant research on driverless vehicles and how this
technology could revolutionize the way people move around and among cities. Specifically,
the purpose of this research is to analyze public opinions on level-5 driverless vehicles and
investigate factors influencing the intention to use this technology. The motivation behind this
thesis is a passion for innovations in the automotive industry of the author, experienced
through an internship done at Daimler AG in the team of electric and autonomous vehicles,
which inspired this work.
The thesis is composed of seven chapters. Following this introduction, Chapter 2 provides a
general overview of driverless vehicles. The chapter begins with the definition and the SAE
International’s categorization of the levels of automation of these vehicles and continues with
their historical overview and current situation. In addition, the second chapter illustrates the
impacts of the Covid-19 pandemic on the development of autonomous driving. Then, the
challenges and the expected benefits of this technology are presented.
Chapter 3 gives an overview of the recent literature and presents the research gap and
questions on which all the research is based. In particular, past surveys and related studies on
public opinions and perceptions about autonomous cars were carefully revised, as well as the
methods for evaluating new technologies. Among them, the Car Technology Acceptance
12
Model has been particularly useful in designing the multiple regression model and its
dependent and independent variables.
Chapter 4 presents the methodology adopted to perform the research. The chapter starts with a
detailed description of the proposed research model and the hypotheses that have been
subsequently tested. It then continues with an explanation of the empirical field study and the
measures used to build the variables of the model.
Building on the above, Chapter 5 explains the results of both the survey and the performed
analysis. The chapter begins with a description of the sample resulted from the responses
collected and an analysis of the reliability of the scales employed in the questionnaire. It
continues with a detailed illustration of how the variables of the model have been created. The
outcomes of the multiple regression analysis, the hypotheses test, and the cluster analysis
close the chapter.
Chapter 6 presents the theoretical contributions of this thesis and its implications, as well as
its limitations and suggestions for further research.
The thesis is concluded in Chapter 7, which summarizes the main findings and insights.
13
2. BACKGROUND ON DRIVERLESS CARS
2.1. DRIVERLESS CARS
2.1.1 Definition
A driverless car is a vehicle capable of perceiving its environment and satisfying the
transportation capacity of a traditional car without human involvement. The human passenger
is neither required to take control of the car nor to be present in the vehicle at all. “An
autonomous car can go anywhere a traditional car goes and does everything that an
experienced human driver does”
1
. To autonomously drive on the street, these cars use a
complex system made of sensors, algorithms, artificial intelligence, machine learning
processes, and processors to execute software. Sensors are located in different parts of the
vehicle and have various functions: Radar sensors are in charge of monitoring nearby vehicles
and objects using radio waves, checking their distance and speed in relation to the vehicle;
Video cameras have the purpose of detecting and interpreting objects in the road such as
traffic lights, road signs, bridges, guardrails, vehicles, pedestrians and cyclists. These cameras
offer a broad picture of the traffic condition around the vehicle, performing the same function
of human drivers’ eyes; Lidar (Light Detection and Ranging) sensors have a function similar
to radar sensors, but they use lasers instead of radio waves. Lidar is used to measure
distances, catch road edges, and identify lane markings. Moreover, they can create 3D images
of the detected objects and build a 360-degree map of the vehicle's surroundings; Ultrasonic
sensors, situated in the wheels, recognize curbs and other vehicles when parking. These
sensors send data to the software that processes the inputs received and delivers commands to
the vehicle’s actuators, which govern all the driving functions such as steering, acceleration,
and braking. Driverless cars' software can perceive vehicles' surroundings in order to
determine the location of the vehicles, and to decide how to behave in different situations.
Artificial intelligence and machine learning systems help the car to learn autonomously to
take the best decision in case of unforeseen events of any kind with minimal response times,
and to handle smartly very complex traffic scenarios in highways and urban environments.
Furthermore, to ensure information exchange and to prevent collisions, driverless cars are
connected to each other and to an internet connection. In this way, the information at vehicles'
disposal increase exponentially, for instance, being able to update routes and maps in real-
time.
1
“What is an Autonomous Car?”. Synopsys
14
2.1.2 Categorization
Several categorizations of autonomy of cars have been proposed in the past years (e.g.
National Highway Traffic Safety Administration (NHTSA), 2013; SAE International, 2014;
Bundesministerium für Verkehr und Digitale Infrastruktur, 2014; Raymann, 2016; Probst,
2016a; SAE International, 2016; Kornhauser, 2017). However, that presented by the Society
of Automotive Engineers (SAE) International in 2016 has become the industry-recognized
standard. SAE categorization identifies six levels of autonomy, from 0 (fully manual) to 5
(fully autonomous) (Figure 2).
Level 0 – No Driving Automation: Level 0 includes all the vehicles that are manually
controlled by a human driver. The automated system may intervene when needed,
without having direct control on the vehicle (e.g. emergency braking system,
conventional cruise control);
Level 1 – Driver Assistance: In addition to conventional cruise control, Level 1
vehicles feature other automated systems that assist the driver, such as the adaptive
cruise control and/or parking sensors. However, the driver is demanded to be attentive
and ready to retake control of the vehicle at any time;
Level 2 – Partial Driving Automation: Equipped with Advanced Driver Assistance
Systems (ADAS), Level 2 vehicles are capable of controlling steering, braking, and
accelerating/decelerating. These vehicles generally have more than one assisted
driving technology that operates at the same time. The driver still cannot turn his
attention away from the driving tasks and must be ready to intervene at any time when
needed. An example is Tesla Autopilot and Cadillac Super Cruise systems;
Level 3 – Conditional Driving Automatic: Level 3 vehicles are able to detect their
surroundings and to make decisions for themselves. These vehicles, using sensors
such as Lidar, take complete control of operating when determined conditions are met
during a route. The driver can do other activities in the meanwhile (e.g. text, watch a
movie). However, the vehicle can solicit the driver to intervene when needed within
some limited time. An example is given by Audi’s AI traffic jam pilot;
Level 4 – High Driving Automation: While the human driver still has the possibility to
manually control the car, Level 4 vehicles can operate in a self-driving mode as they
are capable of intervening in case of system failure. The self-driving mode is allowed
only in limited areas and under specific circumstances, outside of them the vehicles
must interrupt their ride, or the driver must retake control. An example is represented
by robotic taxis or delivery vehicles that operates only in specific locations:
15
Level 5 – Full Driving Automation: Fully autonomous cars do not require human
intervention at all, being able to operate under all circumstances. They are capable of
going anywhere and driving in the same way as an experienced driver. The human
driver's role is completely eliminated, therefore these cars do not have traditional
components such as steering wheel, brake and acceleration pedals, or gear shift
(Figure 1).
Figure 1 - Driverless cars without steering wheel and pedals by GM. Source: Futurism
This research focuses on non-existent level 5 vehicles only being the ones that, by
eliminating the driving task, will completely disrupt the way people move. Indeed, level 5
driverless cars are likely to make the ownership of a traditional car not convenient
anymore, in favor of a service model. In this way, customers can pay for a certain service
(i.e. car as a service), which is likely to be offered at a much lower cost compared to car
ownership. Eventually, car manufacturers could also scale-up from ride-sharing to
"ownership-sharing", where people share the ownership of driverless cars that can go
anywhere they want. Level 5 vehicles require a greater commitment from both the players
of the automotive industry and startups involved in the development of autonomous
driving technology, which must adapt themselves to changing customers’ preferences and
needs, and to completely new business models of transportation. Level 5 driverless
vehicles represent both a major challenge and a new business opportunity for companies.
16
Figure 2 - Levels of Driving Automation. Source: SAE International
2.2. HISTORY OF DRIVERLESS CARS AND CURRENT SITUATION
2.2.1 Historical Overview
The history of the driverless car began 53 years later from the invention of the modern car by
Karl Benz. In 1939, General Motors designed the Futurama exhibit at the New York World’s
Fair which showed their vision of how the world would have looked like in 20 years. This
vision incorporated an automated highway system that would have guided self-driving cars.
By 1958, the concept of the first self-driving car called Firebird II reported in the 1939’s
exhibit was created by the same company (Figure 3).
Figure 3 - The Firebird II concept car by General Motors. Source: Business Insider
17
On the front end of the car sensors called “pick-up coils” were installed, that had the function
of detecting the current flowing through a wire placed in the road. The current could have
been manipulated to command the vehicle to move the steering wheel. Some improvements
came from the Japanese in 1977, which installed a camera system on the vehicle that sent data
to a computer to process images of the streets.
In 1925, Francis Houdina, a former US Army electrical engineer, built the first vehicle
capable of moving without anyone at the steering wheel. To accomplish this goal, he
equipped a 1926 Chandler sedan with a transmitting antenna on the tonneau which collected
radio signals received from a second car that followed it with a transmitter. These signals run
small electric motors that controlled the vehicle’s speed and direction. As reported by the
New York Times, the radio-operated car called American Wonder (Figure 4) was able to turn
on its engine, shift gears, and sound its horn.
Figure 4 - Radio-controlled car American Wonder (Francis Houdina, 1995). Source: Discover Magazine
In 1969, John McCarthy – the founding father of Artificial Intelligence – wrote an essay
entitled “Computer-Controlled Cars”, in which he introduced an “automatic chauffeur”.
Similarly to the modern autonomous cars, this chauffeur was described to be able to navigate
public roads via a "television camera input that uses the same visual input available to the
human driver” [41]. In particular, he imagined cars capable of receiving commands from
users via the chauffeur’s keyboard (e.g. enter destinations, slow down, speed up, organize
stops along the way, find restaurants or bathroom), and autonomously responding to them.
McCarthy’s vision laid the foundation for further research for scholars and engineers after
him.
In 1989, a Ph.D. thesis wrote by Dean Pomerleau, a researcher at Carnegie Mellon University,
defined how neural networks could have permitted a driverless vehicle to betray raw images
from the roads and produce steering controls in real-time. The innovation of Pomerleau lied in
the use of neural nets to teach vehicle to drive, which was a more efficient way compared to
alternative attempts to manually divide “road” and “non-road” images. Six years later, in
1995, Pomerleau and his team from the Robotics Institute at the School of Computer Science
18
at Carnegie Mellon University, developed and put on the road the Navlab self-driving system.
The vehicle equipped with this system was not completely autonomous (speed and braking
were controlled), but the researchers were able to travel 2,797 miles coast-to-coast from
Pittsburgh to San Diego. The journey was called No Hands Across America (Figure 5).
Figure 5 - Pomerleau demonstrates "hands-free" driving. Source: SanFranciscoChronicle
In 2002, DARPA (the research fund of the U.S. Military) launched the first Grand Challenge
addressed to scientists from top research institutions. A $1 million prize would have been
offered to those who built an autonomous vehicle capable of driving a 142-mile course
through the Mojave Desert. The challenges were held in 2004, 2005, and 2007. While in the
first challenge none of the fifteen participants finished the course, in 2005 five vehicles were
able to finish it. In 2007, four cars were capable of finishing the 96km urban area course with
traffic and following traffic laws. In the same years, self-parking systems were introduced as
standard or optional equipment. This was a demonstration that sensors were close to being
capable of facing demanding real-world conditions, like parking in a tight space. The first cars
of being equipped with autonomous parking system have been Toyota’s hybrid Prius and
Lexus in 2003, Ford with its Active Park Assist in 2009, and BMW in 2010. In 2009, Google
secretly began to work on its Self-Driving Project (today known as Waymo), initially led by
the previous director of the Stanford Artificial Intelligence Laboratory and the co-inventor of
Google Street View Sebastian Thrun. After few years of trials, they finally revealed the
autonomous prototype without a steering wheel, a gas pedal, or a brake pedal, able to drive
300,000 miles under computer control without accidents (Figure 6).
19
Figure 6 - Waymo prototype without a steering wheel and pedals. Source: Wikimedia Commons
Other big players in the automotive industry, such as General Motors, BMW, Ford, and
Mercedes-Benz, are working on their self-driving technologies since 2013. While achieving
full autonomous cars is proving to be a hard task, partial autonomy is yet available to
consumers. Tesla, one of the pioneers in the self-driving car market, offers self-driving
systems Autopilot and Full Self Driving on its models since October 2015. These systems
provide the vehicles with partial autonomy (e.g. auto-steering, lane changing, and park
features), but the company is working towards the development of fully autonomous vehicles.
In 2018 Nvidia launched Xavier, the first self-driving chip which integrates artificial
intelligence. The company then unveiled a partnership with Volkswagen aimed at connecting
for the first time Artificial Intelligence to production-ready hardware. This collaboration
represents a huge step forward in the development of driverless cars.
2.2.2 Current Situation
Despite the optimistic view of having “10 million self-driving cars” [8] on the road by 2020
and the huge investments made by car manufacturers and technology companies worldwide,
still there is not a single driverless car available to the public. However, some companies
believe they are on the right way to achieve the full autonomy of vehicles. BMW is
cooperating with Fiat Chrysler Automobiles NV, Intel, and Mobileye NV on level 3 cars, that
should start being tested in 2021. The automaker headquartered in Munich is collaborating
with its main competitor Daimler on the development of driver assistance systems, automated
parking, and level 4 self-driving vehicles. Both companies have declared cooperation to be
"long-term" and "strategic". Mercedes-Benz, in partnership with Bosch, is also developing a
prototype of an autonomous S-Class that will be tested in California. Volkswagen is making
progress on its in-house self-driving technology, which is planned to be launched in 2022.
With its fifth-generation hardware sensor suite, Waymo started to try the self-driving Jaguar I-
Pace - the battery-electric crossover SUV launched by Jaguar two years ago – on public roads
and will soon let its employees try them out. GM Cruise, the self-driving vehicles unit of
20
General Motors, is currently testing 180 self-driving vehicles, which represents the second-
largest autonomous fleet in the world. Nissan launched a revolutionary self-driving system
called ProPilot. The system is integrated with 3D mapping navigation, advanced sensors, and
cameras that even recognize pedestrians' faces, and provide no-hands driving vehicles. The
company is testing ProPilot on its electric-car Leaf. The self-driving startup Zoox, founded in
2014 in Australia, is trying to develop the next generation of mobility-as-a-service in urban
environments by building the world's first fully autonomous fleet. Zoox's prototype vehicle is
fully electric and is not equipped with a steering wheel. Groupe PSA recently partnered with
nuTonomy, an American based self-driving startup and, since 2016, they are developing and
testing their autonomous taxi service around the world, built using Renault
Zoes and Mitsubishi i-MiEVs. Aurora Innovation, the tech company that offers software,
hardware, and data services for self-driving vehicles, is testing its technology on cars of
automakers like Hyundai and Fiat Chrysler Automobiles. Toyota has recently made a huge
investment of 400 million dollars in Pony.ai, the self-driving technology company located in
Silicon Valley, Beijing, and Guangzhou, to accelerate autonomous driving development. The
Swedish carmaker Volvo is working on the production of its next-generation technology,
Highway Pilot, which allows drivers to totally leave the control of the car. Far ahead is the
Chinese internet giant Baidu, which launched its robotaxi service named Apollo Go Robotaxis
in Changsa, Cangzhou, and Beijing (China). The service is currently being offered for free for
two main reasons: collecting data from customers, and win the trust of people. As Zhenyu Li,
Corporate Vice President of Baidu, said: “Baidu Apollo will continue pushing for the
commercial application of autonomous driving. With our technology and platform
advantages, we will contribute more to the development of autonomous driving […]” [61].
Didi Chuxing (DiDi), a ride-hailing app, is willing to develop its robotaxi service, eased by its
great popularity in China. The company, indeed, has more than 550 million users taking
nearly 10 billion passenger trips a year in its human-driven vehicles. Nuro, a company
founded in 2016 by two ex-Google engineers, is working on an autonomous delivery service,
which aims is to allow people to have groceries and goods delivered to their houses by
driverless cars. Tesla CEO Elon Musk at the opening of Shanghai’s annual World Artificial
Intelligence Conference held in July this year said: ”I’m extremely confident that level 5 or
essentially complete autonomy will happen, and I think will happen very quickly” [52]. Musk
seems to be very close to complete level 5 autonomous vehicles this year. The examples
described in this section demonstrate that the commitment to the development of driverless
21
cars is strong and spread throughout the world. However, it will take time for these vehicles to
be ready and for people to trust full autonomy.
2.2.3 Impact of CODIV-19 Pandemic
The Coronavirus pandemic has proved to be another huge obstacle for driverless cars,
delaying their introduction in the market with their original purposes. Many OEMs have had a
huge impact on operations, from production to R&D, experiencing disruption to autonomous
driving development and roll-outs. Companies such as Pony.ai, Cruise, Argo, and Aurora
temporarily stopped vehicle testing and all the operations conducted with human drivers.
Waymo and Ford, that also suspended their testing phase, released data set collected through
their autonomous cars and made them freely available to developers. Data would be used to
find solutions and improvements to increase the robustness of self-driving algorithms and
technologies. Smaller, less-capitalized startup and automakers for which self-driving
technology is not the main business are at risk of leaving the field if they do not adapt
themselves rapidly. Furthermore, the focus of these vehicles is shifting from ride-sharing
service toward goods delivery because of Coronavirus's impact on changing customers'
behaviors. Indeed, there is a reluctance to use shared mobility for hygiene reasons and for fear
of being infected, and greater usage of private vehicles, and this could mean a shift away from
initial Robotaxi's plans. Covid-19, on the other hand, is accelerating a future with autonomous
vehicles used for different purposes. This change towards a reality where working from home,
closed schools and other activities, and local travel restrictions represent the "new normal”,
has made moving products and services without human interactions necessary. Driverless
vehicles could potentially transform the way foods move from restaurants to people's front
door, goods from grocery stores to houses, products from warehouses to storefront, packages
from online retailers to mailboxes, and so on. Covid-19 pandemic, by interrupting the
development of self-driving technology and reducing the purchase of new vehicles from
consumers, has increased the opportunities for autonomous vehicles to be deployed in other
spaces for different purposes. As a matter of fact, driverless vehicles are likely to be primarily
adopted by delivery companies, logistics companies, and foodservice players.
These vehicles have already proved to be essential in the fight against the pandemic. For
instance, in China, they made the transportation of medical supplies and food to health-care
professionals and the public in infected areas easier. They also helped to disinfect hospitals
and public surfaces to reduce the spread of the virus. In Houston, the autonomous delivery
company Nuro made a partnership with a national grocery chain to introduce self-driving
Ricevi informazioni sui nostri servizi, sulle offerte e non perdere news e consigli su università e lavoro.
