SELF-DRIVING CARS WITH MARKOVIAN MODEL-BASED SAFETY ANALYSIS

Abstract

At present, the development of Self-Driving Cars systems has been increasing. The need for man to control all possible scenarios has led to the inclusion of theories such as human perception. This means identifying how the human brain recognizes its environment and translating it into data that a machine can learn and make decisions. For this, great doubts have been generated concerning safety; in the present work, a Markovian model is used, a stochastic method for randomly changing systems. These models show all possible forms and the transitions, the rate of changes, and the probabilities between them. Markovian models can also recognize patterns, make predictions, and learn sequential statistics.

Full text

XXVIII Skup TRENDOVI RAZVOJA: “UNIVERZITETSKO OBRAZOVANJE ZA
PRIVREDU”, Kopaonik, 14 - 17. 02. 2022.
394
Paper No.T4.1-6
11536
SELF-DRIVING CARS WITH MARKOVIAN MODEL-BASED
SAFETY ANALYSIS
Césár Bautista
University Óbuda, Doctoral School of Safety and Security Sciences,
Institute of Next-Technologies, Budapest, Hungary.
cemike.bautista@gmail.com
Abstract: At present, the development of Self-Driving Cars systems has been increasing. The need for man to
control all possible scenarios has led to the inclusion of theories such as human perception. This means identifying
how the human brain recognizes its environment and translating it into data that a machine can learn and make
decisions. For this, great doubts have been generated concerning safety; in the present work, a Markovian model is
used, a stochastic method for randomly changing systems. These models show all possible forms and the transitions,
the rate of changes, and the probabilities between them. Markovian models can also recognize patterns, make
predictions, and learn sequential statistics.
Key Words: Markovian, Perception, Safety, Self-Driving Cars, Stochastic.
1. INTRODUCTION
Self-Driving robotics vehicles are cars or trucks in which human drivers are never required to take control of the
vehicle safely operating. An Internet of Things (IoT) ecosystem consists of web-enabled intelligent devices that use
microprocessors, sensors, and technology to store, processing and act in their environments [1-4].
Internet of Things is also very present in the automotive world. Our cars are becoming more intelligent
(Connected Car) thanks to intelligent sensors that go far beyond calculating a route: they save fuel, they notify
emergency services, our geolocation in case of accident or breakdown, they receive and interpret incidents or safety
notices that affect our trip and communicate it to us in real-time. Through time we have two stages of security inside
the vehicle: passive and active systems [5-9]. Markov defined a way to represent real-world stochastic systems and
processes that encode dependencies and reach a steady state over time. From a mathematical point of view, safety
scenarios are a discrete state space stochastic process without after-effects to be modeled as a Markov Chain.
The paper is organized as follows: In Section 1, the Introduction is given. In Section 2, the Driver Assistance
Technologies are considered. In Section 3, the Self-Driving Robotic Cars are considered. In Section 4, the Intelligent
Cities and Automotive Transportation are presented. In Section 5, the Markovian Model is considered. Conclusions
are given in Section 6 and References.
2. DRIVER ASSISTANCE TECHNOLOGIES
For the safety of the passengers and pedestrians, the designers have established an 'incremental' technological
development. Historically, all the critical points of the vehicle have been developed; today, these advances have
focused on more sophisticated and intelligent systems, so the results are less physically dazzling. For this reason, can
describe two safety systems stages, passive and active safety [10-15].
2.1 Passive Safety
Before World War II, passive safety was established; Hungarian engineer Bela Barenyi designed structures
capable of absorbing impact energy called non-deformable cabins, using the concept of programmed deformation
where the outer structure (chassis) must have high rigidity and gradually decrease towards the inside of the vehicle,
creating a survival cell, e.g., Belt, Headrest, Front airbag, ABS (Antilock braking System), BAS (Brake Assist
System), ESP (Electric Stability Program), etc.
a. Active Safety
The active safety stage has been gaining strength since 2005, based on the implementation of microelectronics
and control systems for vehicle stability, designs capable of correcting driver errors. Concepts of vehicle autonomy
are handled in their fullness, providing the machine with Intelligence to recognize the environment in which the
vehicle circulates and make decisions [16-20].
Since 2012, the development of systems capable of anticipating an eventual accident has begun; the term
"perception" is introduced for a joint work of several sensors for safer navigation and to act independently in an
emergency, e.g., AFS (Adaptive Front Lighting System), TPMS (Tire Pressure Monitoring System), LKS (Lane
Keeping Support), DRL (Daytime Running Lights), eCALL (Emergency CALL), DDM (Driver Drowsiness
Monitoring), SBR (Sit Belt Remembering), EBR Automatic Braking Emergency, etc.

XXVIII Skup TRENDOVI RAZVOJA: “UNIVERZITETSKO OBRAZOVANJE ZA
PRIVREDU”, Kopaonik, 14 - 17. 02. 2022.
395
Figure 1. Passive Safety Systems Figure 2. Active Safety Systems
3. SELF-DRIVING CARS
Self-driving robotics vehicles are AI (Artificial Intelligence)-based cars that will take information about the
environment around them into account before making each decision in which human drivers are never required to take
control of the vehicle safely operating. The main challenge in self-driving cars is the synergy between predictive AI,
imaging technology, sensors, and actuators. The sensors create a 3D picture around the vehicle; the AI interprets the
current vehicle situation, and actuators respond safely; driving a car requires manipulating the direction and the speed.
The imaging technology is overriding; it has become the most significant limiting factor to the wide-scale production
of self-driving cars. Development in sensor technology has historically combined video, radar, ultrasonic sensors,
and Light Detection and Ranging LiDAR [21-26]. This approach, however, has not been successful.
3.1 Perception
Currently, developers use various theories, mechanisms, and techniques to give the vehicle a level of Intelligence
to make decisions. Following this line, they have been based on human behavior and its interpretation of an unexpected
phenomenon, perception. Human perception is defined as the set of processes and activities related to a constant flow
of information, through which we can determine the environment in which we find ourselves, the actions we perform
in it, and our internal state, Figure 3. The perception process in autonomous cars is in charge of recognizing the
vehicle's environment in real-time, using a combination of image acquisition technology (cameras and sensors), 3D
modeling systems, and intelligent algorithms [27-35].
Figure 3. Perception Inside Self-Driving Cars Figure 4. Connecting Traffic Lights
4. INTELLIGENT CITIES AND AUTOMOTIVE TRANSPORTATION
Autonomous vehicle technology enables automobiles to understand the environments in which they operate and
execute safe and efficient commands based on this understanding. Autonomous vehicles can assume decision-making
and operational tasks, enabling drivers to become passengers, entirely disengaged from driving demands. Citizens
engage with innovative city ecosystems in various ways using smartphones and mobile devices and connected cars
and homes. Sceneries of implementation: Connected traffic lights receive data from sensors and cars adjusting light
cadence and timing to respond to real-time traffic, reducing road congestion, Figure 4. Connected cars can
communicate with parking meters and electric vehicle (EV)charging docks and direct drivers to the nearest available
spot. Intelligent garbage cans automatically send data to waste management companies and schedule pick-up as
needed versus a pre-planned schedule. And citizens' smartphone becomes their mobile driver's license and ID card
with digital credentials, which speeds and simplifies city and local government services. Analyze the data: Focus on
two types of analysis: immediate analysis and long-term modeling.
5. MARKOVIAN MODEL

XXVIII Skup TRENDOVI RAZVOJA: “UNIVERZITETSKO OBRAZOVANJE ZA
PRIVREDU”, Kopaonik, 14 - 17. 02. 2022.
396
Markov model is a stochastic process used for estimating categorical time series. The approach has been used in
the transportation domain and other fields such as computer science for speech recognition and social science for life-
course trajectory analysis [33-35]. A Markov model is used to model systems that randomly change states over time.
Further, a first-order Markov model is assumed, i.e., future state at time t+1 only depends on the current state at time
t (and not any previous periods). Assume the activity-travel behavior state of an individual i is represented by X_(i,t)
Where i is an index for individuals and can take values from 1,…,N and t is an index for time and can take values
from 1,…,T. X_(i,t) Can take values from m categories {1,…,m} representing the different activity and travel episodes
that they can pursue. The first-order Markov model is characterized by two parameters θ=(δ,q). δ is a an m dimension
vector of initial probabilities where an element δ_j denotes the probability that the individual starts their day by
conducting activity j at period 0:
δ_j=P(X_(i,1)=j); j∈{1,…,m}
q is an m × m matrix of the transition probabilities, where an element q_(j,h) indicates the probability of switching to
activity ℎ at time t given the individual was pursuing activity j at time t-1.
q_(j,h)=P(X_(i,1)=h├|X_(i,t-1) ┤=j); j,h∈{1,…,m}
q_(j,h) is assumed to be constant over time t. This assumption leads to a homogeneous first-order Markov model.
5.1 Markovian Models
Markov chain - used by systems that are autonomous and have fully observable states. The Hidden Markov model
is used by autonomous systems where the state is partially observable. Markov decision processes - used by controlled
systems with a fully observable state. Partially observable Markov decision processes - used by controlled systems
where the state is partially observable
Every model consists of a structure and parameters that be defined for
the model to be meaningful. The structure of the model defines
dependencies among the various parts of the model. Parameters are values
often, but not necessarily numerical values required by the model.
Parameters are fixed, in which case they constitute model assumptions, or
they may be variable.
5.2 Markov chain
Markov chains are mathematical systems that hop from one "state" (a
situation or set of values) to another. The system is modeled as a sequence
of states and, as time goes by, it moves in-between states with a specific
probability. Since the states are connected, they form a chain.
Figure 5. Markov Chain Example
6. CONCLUSIONS
In Markov model, if can reach all states in the chain, the probability of moving to a particular state will converge
to a single steady value in the long run. In the development of algorithms for autonomous cars, it is not enough to
have the conditions of the car in real-time; it is necessary to consider driving conditions (weather, traffic, etc.) and the
behavior of passengers and pedestrians. Algorithms capable of monitoring these conditions in real-time are generated,
increasing vehicle efficiency, saving resources, adjusting fuel consumption, and directing vehicle control to a safe
point without the need for driver interaction.
3. LITERATURA
[1] César Bautista, Human Perception Inside of a Self-Driving Robotic Car, IPSI Transactions on Advanced Research,
Vol. 17, Issue 2, pp. 50-56, 2021.
[2] Jelena L. Pisarov and Gyula Mester, The Use of Autonomous Vehicles in Transportation, Tehnika – Mašinstvo
(2021), Vol. 76, No. 2, pp. 171-177, DOI: 10.5937/tehnika2102171P, April 2021.
[3] Jelena Pisarov, Gyula Mester, Implementing New Mobility Concepts with Autonomous Self-Driving Robotic
Cars, IPSI Transactions on Advanced Research (TAR), Vol. 17, No. 2, pp. 41-49, ISSN 1820-4511, July 2021.
[4] César Bautista, Gyula Mester, Safety Analysis in Automotive Perception, Full Texts Book of the EJONS 13th
International Conference on Mathematics, Engineering, Natural & Medical Science, Edited by Dr. Mehriban Emek,
Nurlan Akhmetov, pp. 368-378, ISBN: 978-625-7464-40-6, Cappadocia, Turkey, October 26-27, 2021.
[5] Gyula Mester, Jelena Pisarov, Academic Ranking of World Universities 2021, Review of the National Center for
Digitization, ISSN: 1820-0109, Issue: 39, pp. 83-89, 2021.
[6] Albini Attila, Albininé Budavári Edina, Mester Gyula, Adaptation of Bittorrent Technology for Routing Autono-
mous Vehicles, IPSI Transactions on Advanced Research, pp. 28-31, Vol. 17, No. 2, ISSN 1820-4511, July 2021.
[7] Gyula Mester, Jelena Pisarov, Digitalization in Modern Transport of Passengers and Freight, Review of the
National Center for Digitization, ISSN: 1820-0109, Issue: 39, pp. 96-101, 2021.
[8] Jelena Pisarov, Gyula Mester, The Impact of 5G Technology on Life in the 21st Century, IPSI Transactions on
Advanced Research, Vol. 16, No. 2, pp. 11-14, ISSN 1820-4511, IPSI Belgrade, July 2020.

XXVIII Skup TRENDOVI RAZVOJA: “UNIVERZITETSKO OBRAZOVANJE ZA
PRIVREDU”, Kopaonik, 14 - 17. 02. 2022.
397
[9] Jelena Pisarov, Gyula Mester, The Future of Autonomous Vehicles, FME Transactions, Vol. 49, No. 1, pp. 29-35,
DOI: 10.5937/fme2101029P, December 2020.
[10] Jelena Pisarov, Gyula Mester, Rang lista fizičara Srbije, Proceedings of the XXVI Skup Trendovi razvoja:
„Inovacije u modernom obrazovanju” (TREND 2020), pp. 559-562, Kopaonik, Serbia, February 16-19, 2020.
[11] Attila Albini, Gyula Mester, László B. Iantovics, Unified Aspect Search Algorithm, Interdisciplinary Description
of Complex Systems: INDECS, Vol. 17, Issue 1-A, pp. 20-25, 2019.
[12] Jelena Pisarov, Gyula Mester, Programming the mBot Robot in School, Proceedings of the International
Conference and Workshop Mechatronics in Practice and Education, MechEdu 2019, pp. 45-48, Subotica, Serbia,
2019.12.12.
[13] Gyula Mester, Backstepping Control for Hexa-Rotor Microcopter, Acta Technica Corviniensis - Bulletin of
Engineering, Vol. 8, Issue 3, ISSN 2067–3809 Faculty of Engineering Hunedoara, pp. 121-125, 2015.
[14] Gyula Mester, Modeling of Autonomous Hexa-Rotor Microcopter, Proceedings of the IIIrd International
Conference and Workshop Mechatronics in Practice and Education (MechEdu 2015), pp. 88-91, 2015.
[15] Gyula Mester, New Trends in Scientometrics, Proceedings of the 33rd International Scientific Conference Science
in Practice, pp. 22-27, 2015.
[16] Josip Kasac, Vladimir Milic, Josip Stepanic, Gyula Mester, A Computational Approach to Parameter
Identification of Spatially Distributed Nonlinear Systems with Unknown Initial Conditions, 2014 IEEE Symposium
on Robotic Intelligence in Informationally Structured Space (RiiSS), pp. 1-7, Publisher IEEE, 2014.
[17] Gyula Mester, Metode naučne metrike i rangiranja naučnih rezultata, Proceedings of 57th ETRAN Conference,
pp. RO3, 5.1-3, 2013.
[18] Aleksandar Rodic, Gyula Mester, Control of a Quadrotor Flight, Proceed. of the ICIST Confer. pp.61-66, 2013.
[19] Gyula Mester, Univerziteti regiona na Sangajskoj rang listi univerziteta u svetu 2012, Proceedings of the XIX
Skup Trendovi Razvoja, paper No. T1.1-1, pp. 1-5, ISBN: 978-86-78924-95-9, Maribor, Slovenija, 18-21.02.2013.
[20] Aleksandar Rodic, Gyula Mester, Ambientally Aware Bi-Functional Ground-Aerial Robot-Sensor Networked
System for Remote Environmental Surveillance and Monitoring Tasks, Proceedings of the 55th ETRAN Conference,
Section Robotics, Vol. RO2 5, pp. 1-4, Banja Vrućica, Bosnia Herzegovina, Jun 11, 2012.
[21] Gyula Mester, Aleksandar Rodic, Navigation of an Autonomous Outdoor Quadrotor Helicopter, Proceedings of
the 2nd International Conference on Internet Society Technology and Management (ICIST), pp. 259-262, 2012.
[22] Gyula Mester, Felsőoktatási világranglisták 2011, Proceedings of the Informatika a felsőoktatásban 2011
Konferencia, pp. 269-277, ISBN 978-963-473-461-1, Debrecen, Hungary, 2011 augusztus 24-26.
[23] Gyula Mester, The Evaluation of the Impact Factor of the Journal Acta Polytechnica Hungarica, Proceedings of
the TREND 2011, pp. 70-73, ISBN 978-86-7892-323-4, Kopaonik, Serbia, 07-10.03.2011.
[24] Gyula Mester, Aleksandar Rodic, Sensor-Based Intelligent Mobile Robot Navigation in Unknown Environm-
ents, International Journal of Electrical and Computer Engineering Systems, J. J. Strossmayer University of Osijek,
Faculty of Electrical Engineering, Vol. 1, No. 2, pp.1- 8, ISSN: 1847-6996, 2010.12.15.
[25] Gyula Mester, Sensor Based Control of Autonomous Wheeled Mobile Robots, The Ipsi BgD Transactions on
Internet Research, TIR, Vol. 6, Issue 2, pp. 29-34, 2010.
[26] Aleksandar Rodic, Dusko Katic, Gyula Mester, Ambient Intelligent Robot-Sensor Networks for Environmental
Surveillance and Remote Sensing, Proceedings of the IEEE 7th SISY 2009, pp. 39-44, Subotica, Serbia, 2009.09.25.
[27] Gyula Mester, Aleksandar Rodic, Autonomous Locomotion of Humanoid Robots in Presence of Mobile and
Immobile Obstacles, Towards Intelligent Engineering and Information Technology, Vol. 243, pp. 279-293, Springer,
Berlin Heidelberg, 2009.
[28] Gyula Mester, Obstacle Avoidance of Mobile Robots in Unknown Environments, Proceedings of the 5th Intern-
ational Symposium on Intelligent Systems and Informatics, SISY 2007, pp. 123-127, Subotica, Serbia, 2007.08.24.
[29] Gyula Mester, Intelligent Mobile Robot Controller Design, Proceedings of the 10th Intelligent Engineering Sys-
tems, INES 2006, pp. 282-286, DOI: 10.1109/INES.2006.1689384, London, United King-dom, June 26-28, 2006.
[30] Gyula Mester, Modeling of the Control Strategies of Wheeled Mobile Robots, Proceedings of the Kandó
Conference 2006, pp. 1-3, ISBN 963-7154-42-6, Budapest, Hungary, January 12-13, 2006.
[31] Gyula Mester, Neuro-Fuzzy-Genetic Controller Design for Robot Manipulators, Proceedings of the IEEE
IECON'95, International Conference on Industrial Electronics, Control and Instrumentation, Orlando, Vol. 1, pp. 87-
92, DOI 10.1109/ IECON.1995.483338, ISBN 0-7803-3026-9, Florida, USA, 6-10.11.1995.
[32] Gyula Mester, Adaptive Force and Position Control of Rigid Link Flexible-Joint Scara Robots, Proceedings of
the International Conference on Industrial Electronics, 20th Annual Conference of the IEEE, IECON'94, Vol. 3, pp.
1639-1644, ISBN:0-7803-1328-3, DOI:10.1109/IECON.1994.398059, Bologna, Italy, September 5-9,1994.
[33] Gagniuc, Paul A. (2017). Markov Chains: From Theory to Implementation and Experimentation. USA, NJ: John
Wiley & Sons. pp. 1–256. ISBN 978-1-119-38755-8.
[34] S. M. Yang, Y. A. Lin, Development of an Improved Rapidly Exploring Random Trees Algorithm for Static Ob
stacle Avoidance in Autonomous Vehicles Sensors (Basel) 2021 Mar; 21(6): 2244, 2021 Mar 23. DOI: 10.3390/
s21062244
[35] Malik Haris, Jin Hou, Obstacle Detection and Safely Navigate the Autonomous Vehicle from Unexpected
Obstacles on the Driving Lane, Sensors (Basel) 2020 Sep; 20(17): 4719, 2020 Aug 21. DOI: 10.3390/s20174719.