Pengembangan Sistem Autopilot Pada Mobil

Sehingga Dapat Digunakan Untuk Driving Lesson

Dirk Johanes Hendrik Yapsoumilena Joy Zosimus Sakera

Computer Science Department, School of Computer Computer Science Department, School of Computer
Science Science
Bina Nusantara University Malang, Indonesia 65154 Bina Nusantara University Malang, Indonesia 65154
[email protected] @binus.ac

1. Sensors and Cameras:

Autopilot - systems in today's cars have become an
The system must be equipped with sensors and
increasingly developed technology. However, the use of
autopilot systems in driving learning purposes is still cameras to detect the vehicle and surrounding road
limited. Therefore, this study aims to develop an autopilot conditions. This will allow the system to assess the
distance between the vehicle and the surrounding
system in cars so that they can be used for driving
learning purposes. The research methods used in this conditions, and then determine the necessary actions.
study include identification of needs, system design, and
system testing on users. System design is done by
considering various factors such as safety, efficiency, and
comfort. In addition, the system is also equipped with
voice directions that aim to assist the driver in
understanding the situation while driving. System testing
is done by involving drivers who are learning to drive.
Test results show that autopilot systems in cars can help
drivers overcome deficiencies in their driving ability.
With this autopilot system, it is hoped that drivers who
are learning to drive can get a safer and more effective
driving experience.
Keywords — Autopilot, AI, Deep Learning, Driving Lesson,
Car Figure 1. Ultrasonic Sensor [1]

I. INTRODUCTION
Currently, the use of autopilot technology in cars is 2. Artificial Intelligence Algorithms:
growing rapidly. This technology allows the car to be The system must be equipped with artificial
controlled autonomously without driver assistance. In intelligence algorithms that can process information from
addition to improving driving efficiency and safety, the use sensors and cameras. This algorithm can generate actions
of autopilot technology can also be used in driving learning. to reduce vehicle speed if there is a vehicle in front
In this research, we developed an autopilot system for cars approaching or there is an obstruction ahead. This system
can provide information if there are pedestrians,
that can be used in driving lessons. The purpose of this
motorcycles, or cars around using Autonomous Vehicle
research is to create an effective and safe driving learning
platform using autopilot technology. Thus, it is expected that Perception (AVP) technology. AVP is a technology that
can collect information about the environment around
the autopilot system in cars can be an innovative solution in
moving vehicles. The information is then fed into the
improving the quality of driving learning in the future.
learning model to produce the most optimal decision. In
the perception of autonomous vehicles, there are two main
RELATED WORK methods which are Object. Both of these methods mainly
work by using images as their input data. Semantic
To design an autopilot system that can help drivers learn segmentation is a process that identifies each pixel in an
to drive with abilities such as turning on the penny, keeping image and assigns a specific class or label such as people,
bicycles, trees, and so on. This process of semantic
a distance, braking, and so on, here are some steps that can be
taken in designing the system: segmentation is known as the image classification task at
the pixel level [2].
 4. Director System:
The autopilot system must be equipped with a
steering system that can provide instructions and
warnings to the driver when needed. For example,
the system can give instructions to turn on the turn
signal when the vehicle is about to turn or warn if
the distance of the vehicle in front is too close.

In this case, the autopilot system should be equipped

with an easy-to-use and clear interface to help the driver learn
to drive well. The system must also be able to take control
when an ambulance or fire brigade siren is detected. The goal
Figure 2. Sistem deteksi objek disekitar mobil is that drivers who are still learning do not have to panic and
[3] can hand over control to the autopilot system to steer the car
to a safe position. In this feature, an acoustic method is used
that is able to produce accurate detection and distinguish well
3. Automatic Brake System: between emergency vehicle siren sounds and ordinary
The autopilot system must be able to activate the vehicle sounds with high sensitivity [4].
automatic brake system on the vehicle if necessary
to avoid collisions or adjust the vehicle's speed to
road and traffic conditions.

System AI Speaker Head Unit Motor

Users press the learn to drive Gives directions to Displays instructions for using -
feature prepare for steering seat belts, disabling the
position handbrake, stepping on the
gas/brake, etc
Request approval when the driver Tell nearby destinations Open MAP on Head Unit -
is ready to walk as driving practice
Give directions to step on the Give voice instructions Shows an animation of how to If within 3 minutes the
brakes (Ex: Please step on the hit the brakes lever does not move, the
brakes slowly) electric motor will help to
move the brake pedal
down for example
Provides instructions for moving Give voice instructions If within 3 minutes the lever If within 3 minutes the
the transmission lever from P to (Ex: Please move the does not move, the electric lever does not move, the
D lever from P to D) motor will help to move the electric motor will help to
brake pedal down for example move as an example

carried out. This is done by considering various factors

II. METHODOLOGY such as safety, efficiency, and comfort.
The method used in the research "Development of
Autopilot Systems in Cars so that they can be Used for In the design of an autopilot system, adequate design
Driving Learning" is a qualitative method. During the testing must consider safety as a top priority. The system must
phase with users, this study used observation methods and be equipped with safety features that include detection
descriptive analysis to evaluate the driving learning of the vehicle ahead, automatic braking, siren sound
experience using an autopilot system. Here are some steps detection, and collision avoidance. The purpose of these
that need to be done: features is to prevent autopilot failure and potential
traffic accidents. For example, in unusual situations
1. Identify Needs where the color of the vehicle is similar to the
At this stage, user needs for an autopilot system in background color, optical object detection errors can
cars that can be used for driving learning purposes will occur and result in the failure of the autopilot system.
be identified. The autopilot system must be able to This kind of mistake can have serious consequences,
provide feedback and evaluation of the driver's ability including the possibility of a traffic accident. Therefore,
to drive using the autopilot system, so that the driver can careful design and careful safety features need to be
correct deficiencies in their driving ability. implemented to ensure safe and reliable performance of
the autopilot system.
2. System Design
At this stage, the design of an autopilot system on From the above problems, the use of the DIDA
the car that can be used for driving learning purposes is (Integrative Detection With Color Analysis) method to
 solve the problem of optical object detection that can 4. Analysis of Results
occur in situations where the color of vehicle parts is Factors such as complex road conditions, weather
similar to the background color is very necessary. This changes, and other driver behavior can affect the
method combines color analysis with object detection performance of the autopilot system. Therefore, careful
techniques present in the autopilot system. development and testing need to be done to ensure that
the autopilot system functions properly in a variety of
In the DIDA method, color intensity analysis is situations.
carried out on the image produced by the optical sensor
on the autopilot system. Using image processing and In addition, it is also necessary to consider the
pattern recognition algorithms, this method can driver's acceptance and trust in this autopilot
distinguish vehicle objects from the background based technology. Although the technical development results
on significant differences in color intensity. For can be considered good, driver acceptance and trust in
example, if the vehicle has a color that contrasts with the system is also an important factor for its successful
the background, then the DIDA method will be able to use in driving learning purposes.
detect and recognize the vehicle with high accuracy [5].
According to a survey conducted by Reuters/Ipsos,
In addition, the DIDA method also uses an half of the American population believes that cars on
integrative approach, which combines information from Autopilot are riskier than vehicles driven by humans
several other sensors installed in the autopilot system, themselves. Furthermore, two-thirds of respondents
such as radar and lidar sensors. The integration of data stated that they have no interest in buying a car that is
from these sensors can improve the accuracy and fully controlled by a computer [6].
reliability of object detection, especially in complex or
unusual situations.

The comfort factor must also be considered, so the

system must be designed to provide a comfortable
driving experience for the user. For example, the system
can be equipped with a voice direction feature that will
provide direction on how to operate the car from the
moment the driver is in the car until the driver has run
the car on the highway.

3. System Implementation
At this stage, the implementation of the autopilot
system on the car that has been designed is carried out.
This is done by testing and validating the functionality
of the system.

System implementation involves the installation of

the hardware and software needed to run an autopilot Figure 2. Percentage of Drivers Afraid to ride in a
system. Once the installation is complete, the system Fully Self-Driving Vehicle [7].
will be tested to ensure that all components are
functioning properly and can interact well with each
other. From several steps that need to be done above, here are
the results of the needs analysis:
Testing was conducted using scenarios of drivers
learning to drive. In this test, the autopilot system will 5. Supporting technology needs
be activated and the driver will observe how the system Sensors and cameras in cars can be used to monitor
works in different situations, such as when driving on road conditions and traffic situations, as well as identify
the highway, on urban roads, or when encountering an driver behavior in driving, such as speed, visibility, and
emergency. accuracy in maneuvering. These sensors and cameras
will collect data which will then be processed by the AI
Test results will be evaluated and validated to ensure system.
that the autopilot system operates as expected and meets
the needs of drivers who are learning to drive. If The AI system will process the data obtained from
problems or flaws are found, improvements will be sensors and cameras, using pre-programmed
made to improve system performance. All algorithms. This algorithm will identify driver behavior
implementation and testing of autopilot systems will which can be used as an evaluation of the driver's ability
follow standards and procedures established in the to drive. Some examples of technology needed for
automotive industry, including applicable safety autopilot needs such as wanting to give vocal directions
standards and regulations. to the driver to immediately reduce the speed of the car:
 a. Speed Sensor: The speed sensor will provide real- slowing down when entering a corner or signaling when
time information about the speed of the vehicle. needing to change lanes.
b. Voice Recognition System: This system will detect
voice commands given by the driver. Those vocal instructions can be personalized to meet
c. Processing and Controller System: This system will the needs of individual drivers. In addition, the system
analyze data from speed sensors and voice can be programmed to give drivers sound feedback on
commands received. Based on this information, the errors in their driving ability, such as correcting bad
system will generate a command to reduce speed. driving habits or reminding drivers to stay focused on
d. Audio/Speaker System: The in-car audio system or the road.
speakers will issue vocal commands to the driver,
i.e. instructions to reduce speed. 8. Trust analysis of interview results
From interviews with five people, three of them
The way this system works is as follows: expressed confidence in the reliability of the autopilot
system in the need to learn to drive after seeing some of
1. The speed sensor will continuously monitor the the technologies that support it. The latest technology
speed of the vehicle. that will be implemented in this system is designed to
2. The autopilot system will receive voice command keep drivers who are learning to drive safe, especially
input from the driver who is learning to drive. when they are in the car alone, with the help of sensors
3. The voice recognition system will identify the voice installed around the car. However, two others still doubt
command. the reliability of the autopilot system in learning to drive
4. The processing system and controller will analyze because the system relies on computers that have the
data from speed sensors and voice commands. If the potential to experience potentially fatal errors for the
speed of the vehicle exceeds the specified limit, the vehicle and the driver on board. Here are the reasons for
system will generate a vocal command to reduce the the two people who still lack confidence in this
speed. technology:
5. The in-car audio system or speakers will issue vocal
commands to the driver, such as "Reduce speed" or First person:
"Brake now". One respondent who doubted the reliability of the
autopilot system in driving learning was the First
With these vocal directions, drivers who are learning Person. According to him, although the system has been
will be told to reduce vehicle speed as needed. This equipped with various supporting technologies and
helps drivers practice speed reduction more effectively sensors installed around the car, he is still not confident
and safely. in the system's ability to deal with unexpected situations
on the road. For him, humans have better adaptability in
6. Start designing the system dealing with changes in road conditions and other driver
The autopilot system in cars for Driving Lesson can behavior. He worries that complete reliance on the
be designed with safety, efficiency and comfort in mind. autopilot system could reduce the driver's alertness and
The system will involve sensor technology, cameras, skill in dealing with complex situations.
and AI systems to identify weaknesses in the driver's
driving ability. In addition, safety features such as Second Person:
emergency brake, lane departure warning, and adaptive The Second Person also doubted the reliability of the
cruise control can also be added. AI systems can be set autopilot system in driving learning. For him, trust in
up for efficient use in different traffic situations. autopilot technology is still low because of the reliance
Comfort features such as entertainment and temperature on the computers that operate it. He worries about
settings can also be added. potential system errors that could have a serious impact
on driver and vehicle safety. According to him, the
7. Implementation of the system on the vehicle driving learning experience should involve direct
The implementation of an autopilot system in interaction between the driver and instructor who can
vehicles can be done by installing sensor and camera provide feedback directly, which autopilot systems
devices in the car, and connecting them with a pre- cannot offer.
designed AI system.

Sensors and cameras will be placed on several parts 1. RESULT AND CONCLUSION
of the vehicle to collect data on road conditions, traffic
situations, and driver behavior. The data will be 1. Result:
analyzed by an AI system to provide feedback and In this research, we succeeded in developing an
evaluation of the driver's driving ability. autopilot system for cars that can be used in driving
learning. The system is equipped with an easy-to-use
The most important thing is the voice of direction interface and is able to take over control of the car when
for the driver who is learning. For example, the system it detects the sound of an emergency vehicle siren. The
can be programmed to provide vocal instructions on the acoustic method used in the system is able to detect with
steps to take to drive in certain road conditions, such as
 high accuracy and distinguish between the siren sound https://www.weforum.org/agenda/2019/08/self-
of emergency vehicles and ordinary vehicle sounds. driving-vehicles-public-trust/

1. Conclusion: [7] World Economic Forum. (2019, August). Self-

In this research, we have successfully developed an driving vehicles: Public trust [Image]. Retrieved
autopilot system for cars that can be used in driving from
learning. The system has the potential to improve the https://www.weforum.org/agenda/2019/08/self-
quality of driving learning by providing an easy-to-use driving-vehicles-public-trust/
interface capable of providing assistance when needed,
such as in emergency situations. However, although the [8] Surden, H.; Wiliams, M.-A. Technological Opacity,
technical results of developing this autopilot system are Predictability, and Self-Driving Cars; University of
promising, it was found that most of the public still lacks Colorado Law School, Colorado Law Scholarly
trust and acceptance of this technology for personal use. Commons, Colorado Law Faculty Scholarship:
Therefore, in further development, efforts need to be Boulder, CO, USA, 2016.
made to increase public acceptance and trust in this
autopilot technology so that it can be widely applied in
driving learning in the future.

Developing an autopilot system that can be used by

humans for good is a big challenge that must be passed
to achieve good results [8] such as this autopilot
technology to learn to drive. With the results of this
research, it is expected to contribute to the development
of autopilot technology for driving learning as well as
improve our understanding of the factors that influence
public acceptance of this technology. In addition, the
results of this study can also provide direction for
further development in improving the performance of
the autopilot system and strengthening driver
confidence in its use.

III. REFERENCES

[1] TRiABC Official Store (2020). 1048474-00-A

Sensor Parkir PDC untuk Model Tesla S X Sensor
Ultrasonik Parktronik Asli 026303338
104847400A. Diambil dari
https://id.aliexpress.com/item/1005002834085327.
html

[2] H.-H. Jebamikyous, R. Kashef (2022) Autonomous

Vehicles Perception (AVP) Using Deep Learning:
Modeling, Assessment, and Challenges. 10, 10523.

[3] Pratomo FJ (2022). Level 3: Kontrol otonom lebih

menyeluruh. Diambil dari
https://carro.id/blog/berita/mengenal-level-pada-
teknologi-mobil-otonom-yang-canggih/7195/

[4] Tran, V.-T., & Tsai, W.-H. (2020). Acoustic-Based

Emergency Vehicle Detection Using Convolutional
Neural Networks. 8, 75704.

[5] Redmon, J.; Farhadi, A. Yolov3: An incremental

improvement. arXiv 2018, arXiv:1804.02767.

[6] WEF (World Economic Forum). (2019, 19

Agustus). Self-driving vehicles need public trust.
Diperoleh dari