Autonomous vehicles and their impact on road infrastructure and user safety

Full text

Maritime University of Szczecin
Autonomous vehicles and their impact on road
infrastructure and user safety
Anna Skarbek-Żabkin, Marcin Szczepanek
Currently, many car companies have already created
their own self-steering cars. However, they are still not
prepared for road traffic, both urban and extra-urban.
The precursor is the state of Nevada in the USA where
autonomous vehicles were allowed to be tested in road
conditions[5].
In Poland, the city of Jaworzno wants to become a
testing ground for self-steering vehicles. First, appropriate
laws must be created, then Jaworzno will be accurately
metered by a special car collecting data on the road. Tests
can only take place next.
Cars without a driver drive thanks to artificial
intelligence and do not need to interfere with the driving
process. Recognizes traffic signs, picks directions, reacts to
what is happening on the road.
Science-fiction movies have made us accustomed to a
variety of fantasies about autonomous cars. Can you take a
nap while sitting behind the wheel in any such car? Is
driving such a vehicle safe? Who is responsible for the
collision of autonomous cars?
Keywords: autonomous vehicle, vehicle safety, artificial
intelligence
I. AN AUTONOMOUS VEHICLE, WHAT’S THAT?
There are several degrees of autonomy of the car.
Level 0. Means a complete lack of a computer control
system.
Level 1. The vehicle has individual elements to facilitate its
driving, such as: cruise control, parking assistant, or indicator
of the occupied traffic lane. Many ordinary cars are equipped
with this type of facility, and it has to be emphasized that it
does not make them autonomous. This is a kind of first step
towards autonomy.
Level 2. The car is theoretically able to drive alone for a
while, but the system is unreliable to such a degree that it
requires continuous control from the driver. Man has the
responsibility to monitor everything that happens on the road
and be in full readiness to react as a driver.
Level 3. Here the driver is also obliged to constantly watch,
but the control system is a bit more independent. On the
longer road sections it can drive the vehicle by itself.
Level 4. The car drives by itself, without the driver's
participation, but is unable to cope with all conditions (e.g.
weather conditions). Sometimes it may be necessary to drive
manually, therefore a driver is still required in the vehicle.
Level 5. It is only at this level that we can talk about full
autonomy. Passengers can ride in the back seat or the car can
be sent empty - without the risk that it can not cope with the
existing road conditions[1].
Fig.1 - Levels of driving automation [1]
II. ARTIFICAL INTELLIGENCE BEHIND THE WHEEL
OR SYSTEM STRUCTURE
The transport security integration process is currently the most
important aspect. Over the years, research programs were
created, namely Zeus [8] and Gambit [3], whose task was to
improve the safety of the road user, both using the vehicle and
pedestrian. The country that was the first to implement the
Vision Zero program, which assumes that the transport system
does not cause traffic accidents is Sweden. "Vision zero" - this
concept often appears when we mean the description of road
safety in terms of its quality. Autonomous vehicles have a
large share here because they exclude the most unreliable
element of the system which is a human being. Replacement
of the human eye's response to reliable sensors or lasers as in
the case of the LiDAR system.
There were many myths around the so-called "artificial
intelligence" in autonomous cars: from the unpredictability of
control system decisions to the rebellion of machines. It is
worth emphasizing the fact that artificial intelligence is an
ordinary computer program, based on simple, well-known
solutions.
We like when devices make life easier or even work for us.
But do we feel justified worries when the electronic system
was to drive our car? Will it be able to drive more safely than
we do?
In the autonomous car control system, three groups of
elements can be distinguished[6].
1. Input devices or sensors (lidar, cameras, radars), which
are the "eyes" of the system. They collect information about
what is happening around.

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
2
2. Output devices, i.e. actuators. Let's call them "hands" of
our system, because they affect the car directly, controlling the
steering wheel, brakes, etc.
3. Control system. This is the heart, or rather the brain of
the whole system, where all decisions are made, how the car is
to behave. This is the so-called decision system, or in other
words, a regular computer program. It was around it that many
myths were created. Let's verify some of them and see how it's
built.
III. ARTIFICAL HEART OF THE SYSTEM
The decision system may include the following elements:
Algorithms. This is the most common sequence of
mathematical formulas. The algorithm consists of a sequence
of mathematical formulas, shown as equation (1), (2),(3):
𝐴 = 10 × 𝑥 + 15 (1)
𝐵 = 20𝐴 − 2 × 𝑥 (2)
𝐶 = 𝐴 + 𝐵 + 5 (3)
Input data to the algorithm are numerical values, derived
directly from input devices or indirectly from other elements
of the data processing system. The calculations pass
successively through all elements of the algorithm. The
starting information is numerical or logical values.
Fuzzy logic. This way of calculation tries to reflect a
slightly more human way of thinking, related to the
uncertainty of judgments. In this case, the "fuzzy" is a decision
consisting in assigning numbers to specific sets, eg:
𝑥 ∈ 𝐴 𝑖𝑛 80%,
𝑦 ∈ 𝐵 𝑖𝑛 60%,
𝑤𝑖𝑡ℎ ∈ 𝐶 𝑖𝑛 10%.
In this calculation method, a somewhat more human way of
thinking is attempted to reflect the ambiguity of judgments.
Numbers can belong to individual collections to varying
degrees. This allows a better description of a decision that
includes many (often contradictory) factors.
Semantic data sets.
The semantic database stores pairs of information: condition +
application.
𝐴 => 𝐵
𝐵 => 𝐷
𝐶 => 𝐵
It maps the sets of rules that people follow. This allows you
to translate the calculation into a control system, eg: if the
car's speed is higher than the speed limit, reduce the speed.
Input data to the semantic database can be any information
from the other elements of the control system.
Each information consists of a condition and a conclusion
stemming from it. The semantic database stores information,
which allows you to translate the calculations into specific
reactions: if the car's speed is higher than the speed limit,
reduce the speed. We use calculations e.g. from algorithms
and refer them to a semantic database.
Artificial neural networks. The last element is of course the
most exciting. The entire neural network can be theoretically
replaced with one relatively simple mathematical formula
including only addition and multiplication.
Neural is described by a mathematical formula including
simple operations, shown as equation (4):
𝑦 = 𝐴 × 𝑥1 + 𝐵 × 𝑥2 + 𝐶 × 𝑥3 … (4)
Coefficients are selected experimentally. At the entrance to
the network, we provide input data and at the same time show
what values the network should respond to. The network thus
"learns" the correct answers. If we repeat it many times
enough, we will get a ready mathematical formula. You do not
need to develop the formula in a theoretical way; you just need
to have sample data.
Fig. 2 Artificial neural networks [9]
Why then do artificial neural networks get used? Their
advantages lie in the ease of determining the coefficients in
the aforementioned formula. We do not have to calculate them
based on the theoretical considerations. All we have to do is
provide the network with the sample results, and the special
algorithm will select the coefficients itself. At the entrance to
the network, we provide input data and at the same time show
what values the network should respond with. The network
thus "learns" the correct answers. If we repeat it sufficiently
many times, we will get a ready mathematical formula[4].
IV. HUMAN SENSES AND THE SENSES OF MACHINE,
I.E. THE MATTER OF SAFETY
The danger of a car being driven by a computer is completely
different from the dangers of human driving. We are afraid
that the computer program will not include all events that may
occur on the road. What will happen if a truck appears in front
of us, whose colour blends with the colour of clouds in the
background. Will the computer program recognize its
contour?
It must be stressed, however, that driving by a man is also
not free from dangers. The ability to test the operation of the
system under any conditions definitely tilts the safety scales in
favour of autonomous cars.
On the driving license course, one learns the basics of
driving a vehicle. As a result, he can behave in many standard
situations that can happen in practice. One can not predict how
he will behave in an emergency situation, and it is these
situations that carry the greatest danger of an accident. Despite
this, we entrust the driver with the task of driving a car. The
situation is quite different in the case of a computer control
system. In fact, it is not a simple matter to prepare a program
that will be able to control the vehicle. But when it is ready,
we do not entrust him with driving yet. Now we can check
how it will behave in a variety of situations that ordinary
drivers have not even dreamed of.
We have this freedom of testing because the control system
operates on a computer, so on the computer we can also

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
3
simulate individual dangerous situations. We can make a
simulation and see how the decision system behaves. To be
sure, let's check it out for 20 million cases, including various
lighting, lacquer shades and other factors. It will take a few
seconds for the computer and we will know exactly what to
expect from the tested program. We can simulate millions of
dangerous situations and even breakdowns of the car and
adjust our control program in such a way that it can faultlessly
handle the vehicle in all conditions. These possibilities are
impossible even for long-term professional drivers. Artificial
control system can be safer than the best human driver, which
is influenced by various factors, including fatigue, stress,
distraction or bravado. These factors are completely alien to
the control systems.
V. RESPONSIBILITY OF MAN OR MACHINE?
Who then drives an autonomous car? The correct answer may
sound a bit perverse, but it is a man and only him. The
computer only performs the simple tasks to which it was
prepared, but it is the human mind that must program it. In this
way man manages the vehicle indirectly - through the program
he develops.
Autonomous cars will be as safe as well-educated are the
specialists who build them. That is why investing in the
research is so important. Before us is a long and incredibly
interesting road to develop machines that will relieve us of
many mundane activities of everyday life.
Communication between vehicles on the road is the
foundation of safety. V2V (Vehicle To Vehicle) and V2I
(Vehicle To Infrastructure) systems provide improved road
traffic, driving conditions as well as mobility and safety. V2V
warns of the danger of cullet, violent braking of the car ahead
of us or changing the lane by a car moving in parallel. The
relays of vehicles equipped with V2V technology (vehicle-to-
vehicle) send information about location, direction of travel
and speed, while other vehicles automatically analyze this data
and inform drivers about the risk of collision. The system can
also inform about traffic, which in turn significantly translates
into increased driving fluidity, reduced fuel consumption and
exhaust emissions[2].
Fig.3Vehicle to vehicle [10]
Example of operation of V2V technology:
The system warns the driver about a car that has stopped on
his driving lane as a result of a breakdown. Even if the vehicle
is not in the lane of the car and does not pose a direct threat,
the system informs the driver of the appropriate symbols even
before making eye contact, so that he knows about a
potentially dangerous situation and is sensitive to, for
example, the presence of pedestrians on the roadside.
Example of operation of V2I technology:
Vehicle-to-Infrastructure (V2I) is the next generation of
Intelligent Transportation Systems (ITS). V2I technologies
capture vehicle-generated traffic data, wirelessly providing
information such as advisories from the infrastructure to the
vehicle that inform the driver of safety, mobility, or
environment-related conditions. State and local agencies are
likely to install V2I infrastructure alongside or integrated with
existing ITS equipment. Because of this, the majority of V2I
deployments may qualify for similar federal-aid programs as
ITS deployments, if the deploying agency meets certain
eligibility requirements.
Adjustment of road infrastructure
Autonomous vehicles are adapted to the existing road
infrastructure. One of the most important is to reflect the
infrastructure in the navigation device. Systems responsible
for vehicle steering, using sensors and cameras are able to read
road signs. However, they need detailed data about the space
they are moving on to compare the existing state with the state
stored in the navigation database. TomTom is the first
company to start creating maps for three-dimensional
navigation. They created the RoadDNA system and HAD
(Highly Automated Driving) maps, which enable vehicle
location, both at high speed and in difficult weather
conditions. As a result, a detailed and optimized route image is
provided so that the vehicle can compare it with the image
from cameras and sensors[7]
Fig. 4 Real view and view from RoadDNA map [7]
During the CES fair (Consumer Electronics Show) of the
world's largest electronics trade fair and new technologies in
Las Vegas, TomTom presented the latest solutions for
autonomous vehicles. The first of these is TomTom
AutoStream - a completely new map service that allows
vehicles to build the horizon before the planned route, by
streaming the latest map data from the TomTom cloud.
Another new feature is TomTom MotionQ, a unique driving
concept designed to ensure the comfort of passengers using
self-propelled vehicles[7].
VI. SUMMARY
The rapid development of autonomous cars technology and
the interest of companies and corporations in the creation of

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
4
their own vehicles of this type indicates that intelligent
technologies on the road are the future of transport. It is a
chance to improve road safety and turn towards ecological
thinking about transport. In addition, autonomous mobility
improves the mobility of people with disabilities and older
people. The most serious problem of common infrastructure
for conventional and autonomous vehicles is to solve legal
problems. This solution requires the development of common
rules for self-aware vehicles for all countries, making ethical
decisions, and who is at fault for causing an accident: vehicle
manufacturer, system manufacturer used in the vehicle, or
driver - passenger.
The commencement of cooperation at the government level
in the field of road automation in Poland was announced by
the Minister of Infrastructure and Construction Andrzej
Adamczyk and the Minister of Energy Krzysztof Tchórzewski
during the international conference "Autonomous future of the
road transport". It was organized by the Motor Transport
Institute in Warsaw. Over four hundred representatives of the
automotive industry and the business world as well as
scientists have registered to participate in the meeting. Among
them are guests from Israel, the United States, Germany and
Belarus.
First, the autonomous cars will appear in enclosed areas, i.e.
it will be testing under special supervision. However, their full
commercial use is planned for around 2025. Then all
guidelines for self-steering cars will be created.
REFERENCES
[1] Automated and Autonomous Driving, OECD/ITF, European
Parliament ‘Automated vehicle in the EU’, Briefing, January
2016,
[2] Delphi V2E – nowy system komunikacji samochodów
autonomicznych http://technowinki.onet.pl,
[3] Gambit Project http”//www.gambit.fril.org.pl, Accessed on:
Jun. 15, 2018
[4] Pobocha B., Rozwiązania dla autonomicznych samochodów
przyszłością branży nawigacyjnej, Nawigacja samochodowa
& Connected Car, 2015,
[5] Przybysz K., Infrastruktura przystosowana do pojazdów
autonomicznych, KOKONAT, 2016, Kraków,
[6] Skarbek-Żabkin A., Szczepański T., Kto kieruje pojazdem
autonomicznym’ Czysta Energia Plus, Elektromobilność,
1/2017,
[7] TomTom ‘RoadDNA’ Raport and Scatable loocation
technology,
[8] Zeus Project http://www.e-zeus.eu, Accessed on: Jun. 16,
2018
[9] Mohamed A.Shahin, State-of-the-art review of some
artificial intelligence applications in pile foundations,
Geoscience Frontiers, Volume 7, Issue 1, January 2016,
Pages 33-44
[10] Q Xu, T Mak, J Ko, R Sengupta , Vehicle-to-vehicle safety
messaging in DSRC, VANET ’04 Proceedings of the 1 st
ACM international workshop on Vehicular ad hoc networks,
pages 19-28
Skarbek-Żabkin A. M.Sc. Eng. –
Graduate of Environmental Engineering at the
Warsaw University of Technology and
Interdepartmental Studies of Environmental
Protection at the University of Warsaw,
specializing in Environmental Protection Law.
Since 2011, she has been professionally connected
with the Motor Transport Institute. He is a
member of the Polish Scientific Society of
Recycling, Polish Scientific and Technical Society
of Exploitation and the Innovative Scientists Club.
Specializes, among others in issues related to
environmental protection, unconventional drives of motor vehicles,
alternative fuels as well as electric and autonomous vehicles. She took
an active part in the implementation of the European HIT-2-Corridors
project - Hydrogen as fuel, projects related to energy recuperation
research and modeling of engine processes. Author and co-author of
several scientific publications in national and international magazines in
the field of environmental protection, electromobility and transport.
Since 2016, a PhD student at the Maritime University of Szczecin.
Szczepanek M. PhD, DSc, Eng.
Employed by Maritime University of Szczecin,
Appointed on position of charge Deputy Dean of
Faculty of Marine Engineering since 2016. Area
of scientific research: energy audits of vessels,
alternative fuels, scrapping and recycling of
technical floating facilities. Author and co-author
of several scientific publications in national and
international magazines in the field of
environmental protection, and transport.