SINERGI Vol. 29, No. 3, October 2025: 587-598
http://publikasi.mercubuana.ac.id/index.php/sinergi
http://doi.org/10.22441/sinergi.2025.3.003

Driver assistance collision warning system using a LIDAR
sensor with kinematics and perception algorithms
Willi Immanuel Susanto1, Henry Nasution2*, Tanika Dewi Sofianti1

1Department of Mechanical Engineering, Faculty of Engineering and Information Technology, Swiss German University,

Indonesia
Renewable Energy Engineering Technology, Faculty of Industrial Technology, Bung Hatta University, Indonesia

2

Abstract
Road accidents remained a significant global concern, causing loss
of life and economic damage. To mitigate this issue, the automotive
industry has increasingly invested in Advanced Driver Assistance
Systems to enhance vehicle safety. This research presented a Driver
Assistance Collision Warning System that incorporated kinematics
and perception algorithms to improve collision prevention. The
system utilized a LIDAR sensor to capture real-time data regarding
the distance to the vehicle in front of it. This data was integrated with
an Arduino microcontroller to compute the relative speed and time of
collision. Upon detecting a collision risk, the system triggered a
warning mechanism, which included an audible alert provided by a
buzzer and a visual warning displayed on the head-up display. The
system integrated kinematics algorithms, which processed sensorread values to generate real-time decisions utilizing a specific
threshold time to collision, and perception algorithms relied on Fuzzy
Logic to handle uncertainty and improve accuracy. Validation was
conducted through integration, system, and acceptance testing,
demonstrating reliable synchronization of algorithms and accurate
operation in real-world environments. The results showed that the
system achieved a collision risk detection accuracy of ±5 cm within
five different environmental factors. These findings confirmed the
system's potential as a reliable solution for real-world collision
prevention.

Keywords:
Collision;
Driver Assistance;
Kinematics Algorithm;
LIDAR;
Perception Algorithm;
Article History:
Received: August 2, 2024
Revised: January 23, 2025
Accepted: February 5, 2025
Published: September 1, 2025
Corresponding Author:
Henry Nasution,
Renewable Energy Engineering
Technology, Faculty of Industrial
Technology, Bung Hatta
University, Indonesia
Email:
henrynasution@bunghatta.ac.id

This is an open-access article under the CC BY-SA license.

INTRODUCTION
Road accidents are an inevitability. The
World Health Organization reports that over 3,000
individuals perish in vehicle accidents daily [1].
The National Transportation Safety Committee
(KNKT) of the Republic of Indonesia reports that
the total number of road accident cases in
Indonesia was 100,028 in 2020. The situation
worsened in 2021, with road accidents increasing
by approximately 4% [2]. Road traffic injuries
cause substantial economic harm to individuals,
their families, and the nation. These losses stem
from treatment expenses and diminished output

for individuals who are killed or incapacitated due
to their injuries, as well as for family members who
must forego a job or education to assist the
injured. According to reports, material losses in
2021 are likely to reach IDR 247 billion due to
monetary impacts [2].
A variety of technologies have been
developed in the area of Advanced Driver
Assistance Systems (ADAS) that offer safety
features designed to enhance driver safety and
facilitate a more comfortable driving experience
and reduce road accidents. Research by the
Traffic Safety Researcher indicates that findings

W. I. Susanto et al., Driver assistance collision warning system using a LIDAR sensor …

587

SINERGI Vol. 29, No. 3, October 2025: 587-598

from police-reported crash analyses are
converging, suggesting that Vulnerable Road
User ADAS decreases pedestrian crashes by 13%
to 27% [3]. Theoretically, the adoption of ADAS in
automobiles is supposed to reduce the frequency
of road accidents. Notwithstanding, ADAS
systems are presently only integrated into
contemporary high-end vehicles. The car's price
exhibits a linear correlation with the types of
technologies used in safety systems. These
findings are inconsistent with consumer demand,
as the safety system is crucial for all customer
segments.
In recent years, vehicle safety systems
have been a significant focus of research, with
numerous recommendations for methodologies
put forth. Numerous studies have been conducted
on this subject; the following are several notable
studies that address the issue.
H. Kunto D. A. conducted a study on a
vehicle anti-collision system utilizing Arduino Uno.
This system employs object detection sensors
(ultrasonic and laser range finder) as inputs to a
warning system output (buzzer and LED) through
kinematic logic [4]. In another study, Yuan, Yuwei
Lu, and Qi Wang investigate car driving assistance
based on the driver's facial positions, utilizing a
dataset constructed by the researchers.
Technology utilizes machine learning algorithms
that require comprehensive pre-trained models to
forecast collision risks based on historical driver
behavior [5]. While both H. Kunto’s kinematicbased system and Yuan, Yuwei Lu, and Qi Wang's
perception-based approach provide valuable
insights, each has limitations. Kunto’s method
lacks adaptability to environmental changes, and
Yuan's system, though adaptable, is hindered by
its reliance on extensive pre-trained models,
which compromise real-time responsiveness. To
address these issues, the planned research aims
to develop a hybrid system that combines the
strengths of both kinematic and perception-based
approaches. This new system aims to deliver realtime accuracy and adapt to changing conditions,
thereby enhancing collision prevention.
This research aims to develop a brandagnostic Driver Assistance Collision Warning
System that ensures universal compatibility
across vehicle brands and segments through
Arduino, while seamlessly integrating LIDARbased kinematics algorithms for real-time
decision-making with AI-driven perception
algorithms utilizing fuzzy logic to enhance collision
prediction accuracy and adaptability in diverse
driving conditions. The kinematics-based method
uses sensor-read values as input to produce a
decision output. The perception-based algorithm

588

employs Artificial Intelligence (AI) logic to analyze
rule-based data from the fuzzy system. Rulebased systems are one of the stages of an AI
system, where a computer uses rules [6].
The
system
integrates
kinematics
algorithms, which process sensor readings to
generate real-time decisions, and perception
algorithms, which employ fuzzy logic to handle
uncertainty and enhance accuracy. This
integration ensures intelligent collision reduction,
adaptable across vehicle brands and segments.
METHOD
The research paper presents a theoretical
framework that establishes the development of the
Driver Assistance Collision Warning System.
Driver Assistance System reduces exposure to
hazardous situations and enhances driving
comfort by providing warnings or automating
dynamic driving tasks [7]. This system aims to
intelligently minimize collisions by integrating
kinematic and perception algorithms. The
research methodology to be utilized for the
Vehicle Collision Warning System is the V-model
project milestone approach. The V-model
approach to development, well-established in the
automotive industry, is subject to high regulations
imposed by the requirement for compliance with
standards [8]. Compared to the other method, the
V-model provides more proper handling for
support software integration [9]. The V-model
methodology divides the development phase into
design, implementation, integration, and system
testing.
V-Model Project
The letter “V” symbolizes the development
flow, with the left side indicating requirements and
specifications, while the right side represents
verification processes. The horizontal connection
between the left V side and the right V side
signifies that verification must adhere to the
requirements.
The V-model begins with the requirement
stage, also known as the pre-development stage,
where the system's requirements and architecture
are established. Moving down the left side of the
“V”, the main development stage involves
Hardware
(H/W)
and
Software
(S/W)
development, following three phases: component
development, implementation with unit analysis,
and final integration. After development, the
process shifts to the right side of the “V” for
System Verification, ensuring validation and
testing align with the defined requirements. Figure
1 illustrates the detailed flow.

W. I. Susanto et al., Driver assistance collision warning system using a LIDAR sensor …

p-ISSN: 1410-2331 e-ISSN: 2460-1217

Table 1. Warning Symbol
Function

Illumination
Color

Emergency
Brake

Warning for
emergency braking
requirement upon
detection of an
incoming front
collision.

Yellow

Brake

Warning for taking
braking action: light
brake, moderate
brake, significant
brake

Green,
Yellow, Red

Item

Control
Symbol

Figure 1. Research Method using V-Model
Pre-development Stage
The pre-development stage defines the
essential system requirements in two phases:
Requirement and Architecture Development. The
driver assistance system selects three primary
variables: distance to the front vehicle (d), relative
speed (vr), and time-to-collision (TTC). The
sensor measures distance in centimetres and is
integrated with an Arduino microcontroller, which
calculates relative speed to determine the TTC.
The TTC notion denotes the time it takes for the
front end of the following vehicle to reach the rear
end of the leading vehicle, assuming both vehicles
proceed at their current speeds and on the same
path [10]. The time-to-collision combines the
spatial distances with the (relative) velocities to
quantify the ’distance’ to a collision [11]. Equation
(1) and (2) shows the Relative Speed and the TTC
formula.
𝑣𝑟

𝑑𝐵 − 𝑑𝐴
𝑡𝐵 − 𝑡𝐴

(1)

𝑑
(2)
𝑣𝑟
Furthermore, the second requirement is the
warning output, categorized into two types:
audible warnings, which deliver immediate
notifications by sound, and visual warnings. The
UN ECE guarantees that these laws are both
universally
implemented
and
inclusive,
encouraging a global approach to road safety.
Standards and conventions developed in UNECE
are used worldwide [12].
The audible warning device shall emit a
continuous and uniform sound; its acoustic
spectrum shall not vary substantially during its
operation [13]. At the same time, three visual
indication categories are mentioned in the UN
ECE Regulation No. 121: Control, Tell-tale, and
Indicator [14]. The study incorporates an electric
buzzer for audible warnings and LEDs for the
visual warning category. The details of the
warning symbol are outlined in Table 1.
𝑇𝑇𝐶 =

Table 2. Kinematics Algorithm Guide
Time-toCollision
(sec)

Actuator
Emergency Brake
LED

Buzzer

>3

Off

Off

2 > TTC ≤ 3

Amber

Off

≤2

Amber

On

Table 2 shows the system’s decision based
on the 3-second rule and speed limit, which serves
the final requirement in the pre-development
stage. The vehicle collision warning system must
adjust its modifications in guidance based on
varying conditions. The driver assistance system
that has been developed will be implemented as a
kinematics and perception algorithm used to make
decisions based on facts.
The kinematics algorithm follows a strict
true/false approach. It adheres to the 3-second
rule as advised by the educational movement from
the Toll Road Regulatory Agency of the Ministry of
Public Works and Public Housing [2]. The older
recommendation is the following 2-second rule.
However, based on a study by highway engineers,
states and traffic safety organizations have more
recently referred to a 3-second rule [15].
Meanwhile, the perception algorithm
applies Fuzzy Logic to suggest braking actions
based on distance and relative speed. It classifies
inputs into four levels: Significant, Moderate, Light,
and Safe, determining the appropriate braking
response.
The
final
pre-development
stage
progresses to architecture development through
the components block diagram, as shown in
Figure 2.
The system employs the LIDAR sensor to
measure the distance to the front vehicle. LIDAR
is a distance sensor that is useful for the
development of ADAS and autonomous driving
[16, 17]. The vehicle's relative speed and Time-toCollision are derived from calculations performed
by the Arduino Mega 2560.

W. I. Susanto et al., Driver assistance collision warning system using a LIDAR sensor …

589

SINERGI Vol. 29, No. 3, October 2025: 587-598

Figure 2. Collision Warning System Block Diagram
It operates at 5 V and is easy to use, not least
because several electronic components operate
at the same 5 V [18]. To assist the driver with
critical safety information, the system provides
safety warnings through two actuators: an LEDbased head-up display (HUD) for visual alerts and
a buzzer for auditory signals to grab the driver’s
attention.
Calibration, Testing, and Validation
A critical phase of unit analysis and testing
is calibration, which involves evaluating and
validating each component.
Calibration is
essential to guarantee that the data collected is
not only crisp but also accurately represented.
Calibration enables the systems to be adjusted for
natural driving habits, hence enhancing customer
acceptance of driver assistance systems [19]. The
calibration procedure consists of two phases:
confirming that the LED and buzzer react
appropriately to the microcontroller's inputs and
ensuring the LIDAR sensor accurately measures
the actual distance.
Additionally, testing and validation ensure
that the methodology, data, and results align with
the research objectives. According to the V-Model
project
milestone,
the
validation
phase
necessitates verifying the accuracy of the
temporal configuration by comprehensive testing
of the implementation on the target [4]. The phase
involves three testing methods: integration testing,
system testing, and acceptance testing.
RESULTS AND DISCUSSION
The current phase has progressed to the
primary development stage of the V-model, which
provides an in-depth examination of the structured
steps involved in development, calibration, and
testing. These discussions are based on the
design concepts established during the predevelopment phase.
Hardware Development
In terms of hardware, this necessitates
workable specifications that are feasible. A
workable specification is needed to start
producing a prototype for the design, ensuring that
the design specification is well-drafted [20]. The

590

wiring diagram serves as a workable specification,
detailing all electrical connections, including cable
arrangement, components, and connection points.
Figure 3 illustrates the system’s wiring diagram.
The wiring diagram built with Fritzing
software serves as an essential blueprint for the
assembly
and
integration
of
hardware
components. The hardware assembly displayed in
Figure 4 is executed by referencing the blueprint
specifications in the wiring diagram.
Software Development
Software development requires the creation
of a calibration program to synchronize the
hardware's output with established standards. The
calibration program is organized based on an
integrated system of sensors and actuators.
The actuator calibration utilizes a calibration
approach that involves a distinctive Arduino
program, adhering to the flowchart in Figure 5. The
result of the actuator calibration is displayed in
Table 3.

Figure 3. Wiring Diagram of Warning System

Figure 4. Vehicle Collision Warning System Assy
Table 3. Calibration Result for LIDAR Sensor
Actual
Range

System
Read

(cm)

(cm)

55

Standard
(cm)

Result
(cm)

56~59

± 5 cm

Max +4 cm

90

93~95

± 5 cm

Max +5 cm

280

284~285

± 5 cm

Max +5 cm

565

566~568

± 5 cm

Max +3 cm

1520

1522~1524

± 5 cm

Max +4 cm

Remark
Within
Spec
Within
Spec
Within
Spec
Within
Spec
Within
Spec

W. I. Susanto et al., Driver assistance collision warning system using a LIDAR sensor …

p-ISSN: 1410-2331 e-ISSN: 2460-1217

Table 5. Kinematics Algorithm Decision

Figure 5. Actuator Calibration Flow Procedure
The results of the sensor calibration are
displayed in Table 4. Sensor calibration ensures
the accuracy of the Garmin LIDAR Lite v3 by
comparing values acquired from a manual
distance measurement using a tape measure with
the distance output displayed in the Arduino serial
monitor.
Integration Build
This phase involves integrating several
software and hardware components to create a
unified system. The integration of the system
entails three steps in the decision-making
process: the kinematics algorithm, the perception
algorithm, and the comprehensive system that
integrates kinematics and perception algorithms.
The Kinematics algorithm approaches the
specified target following the 3-second rule
guidance. The Arduino program has been set up
with three if-conditions, as shown in Table 5, while
also considering the computation time required for
measuring distance using LIDAR.
Table 4. Actuators Calibration Results
Actuator
Buzzer
LED Emergency
Brake
LED
Green
Retarder
LED
Amber
Retarder
LED
Red
Retarder

Step
1

Step
2

Step
3

Step
4

Step
5

ON

OFF

OFF

OFF

OFF

OFF

ON

OFF

OFF

OFF

OFF

OFF

ON

OFF

OFF

OFF

OFF

OFF

ON

OFF

OFF

OFF

OFF

OFF

ON

Time-toCollision
(sec)

Interval
(sec)

Actuator
Emergency Brake
LED

Buzzer

>2

1

Off

Off

1 > TTC ≤ 2

1

Amber

Off

≤1

1

Amber

On

The perception algorithm relies on Fuzzy
Logic. As more scenarios necessitate decisions
that cannot be resolved with a mere yes or no
response, the use of fuzzy logic to facilitate
decision-making becomes increasingly essential
[21]. The development of Fuzzy Logic will be
executed utilizing MATLAB. To establish a fuzzy
logic system that offers suggestions for braking
decisions based on the distance to the front
vehicle and the relative speed, the following is a
comprehensive and systematic approach to
constructing the fuzzy system:
First Step: Address the Inputs and Outputs
The development begins with the
identification of the variables. The model's inputs
are the distance to the front vehicle and the
relative speed. The result displays the Brake
Suggestion, specifying the recommended braking
action. A crisp value within a specified range
must be present in every variable. Table 6
presents comprehensive data regarding the
precise crisp values for each variable.
Second Step: Fuzzification
The fuzzification process subsequently
transforms crisp value inputs into fuzzy inputs. The
condition is accomplished by assigning every
variable of input to a collection of linguistic
concepts, each indicated by a fuzzy membership
function (MF). The MF type is characterized by the
application of Trapezoidal and Triangular shapes,
attributed to its widespread use and good
performance. Nasution's 2011 research indicates
that the type is simple, providing good controller
performance and being easy to handle [22]. Table
7 illustrates the membership function of each
variable.
Subsequently, the crisp value was
generated in line with the given requirements and
is currently being incorporated into the fuzzy
system built with MATLAB.
Table 6. Input-Output of Fuzzy Variables
Variable
Distance to Front
Vehicle
Relative Speed
Brake Suggestion

Specification

Range

0–4m

0 – 4000

0 – 4 m/s
100% brake
application max

0 – 4000

W. I. Susanto et al., Driver assistance collision warning system using a LIDAR sensor …

0 – 100

591

SINERGI Vol. 29, No. 3, October 2025: 587-598

Table 7. Fuzzy System Membership Function
Variable

Distance
(to Front
Vehicle)

Relative
Speed

Brake
Suggest

Membership
Function

Type

Parameters

Close

Trapezoidal

[0 0 500
1000]

Fair

Triangular

[500 1500
2500]

Far

Trapezoidal

[2000 2500
4000 4000]

Slow

Trapezoidal

[0 0 500
1000]

Medium

Triangular

[500 1500
2500]

Fast

Trapezoidal

[2000 2500
4000 4000]

Safe

Triangular

[0 15 30]

Light

Triangular

[20 35 50]

Moderate

Triangular

[40 55 70]

Significant

Triangular

[60 80 100]

Figure 6 illustrates the fuzzy system’s input and
output developed in MATLAB.
Third Step: Define Rule
The rules consist of a series of IF-THEN
statements that establish a logical relationship
between the inputs and the outputs of the fuzzy
system. With two variables, each containing three
membership functions, the total number of viable
rules amounts to nine. The decision is determined
based on the initial forecast after evaluating all
possible rules. The system's Fuzzy Rules are
illustrated in Figure 7.
Fourth Step: Fuzzy Inference Method (Engine)
The MATLAB Fuzzy Logic Designer offers
two types of inference engines: the Sugeno and
Mamdani types. The Mamdani model will be
employed in the collision warning system due to
its enhanced reliability in producing accurate
outcomes. A previous study by Mateichyk et al. on
the energy efficiency of vehicles has shown that
the Mamdani-type fuzzy system yields better
results compared to the Sugeno-type [23].
Fifth Step: Defuzzification
Defuzzification
is
the
process
of
transforming fuzzy outputs into exact results. The
collision warning system will utilize the centroid
approach for defuzzification. As previously stated
in the research, the centroid method is one of the
most widely used methods in engineering
applications, where membership values are
treated as weights to produce a balanced and
representative crisp output [24].

Figure 6. Fuzzy System Membership Function

Figure 7. Fuzzy System Rules Mapping
In the centroid method, the fuzzified value,
dCA(C), is defined as the value within the range of
variable z for which the area under the graph of
the membership function C is divided into two
equal subareas. For the discrete case, in which C
is defined as a finite universal set [z1,z2,…,zn], the
formula is presented in (3) [25].
𝑑𝐶𝐴 (𝐶 ) =

∑𝑛𝑘=1 𝑐(𝑥 ) 𝑧𝑘
𝑘
∑𝑛𝑘=1 𝑐(𝑧 )

(3)

𝑘

Upon finalizing the construction of the fuzzy
system from the first to the fifth step, attention now
turns to the final phase of testing and optimization.
Fisrt Final Phase: Implementation and Testing
The first final step of implementation and
testing demonstrates the system's operational
use, which includes the development of a
graphical user interface (GUI) using MATLAB’s
rule inference capabilities.
In the MATLAB GUI, the crisp input for
distance is defined as 2100, corresponding to
2100 cm, and the relative speed is defined as 700,
equating to 700 cm/s. The output from the fuzzy
system is 35, indicating a requirement of 35%
brake application. To determine the system's
functionality, a manual calculation of the fuzzy
decision must be performed, as in Figure 8.

Figure 8. Manual Fuzzification

592

W. I. Susanto et al., Driver assistance collision warning system using a LIDAR sensor …

p-ISSN: 1410-2331 e-ISSN: 2460-1217

mean determines the range of each variable for
the three membership functions.
Brake Safe: zA = 15
Brake Light: zB= 35
Brake Moderate: zC = 55

Figure 9. Rule Evaluation
a. Fuzzification
Utilize the crisp inputs d1 and vr1 to
calculate their respective degrees of membership
(DOM) within the relevant fuzzy sets.
Crisp input: d1 = 2100; vr1 = 700
Membership function: Close (dc); Fair/Moderate
(dm); Far (df); Slow (vrs); Medium (vrm); Fast (vrf)
Figure 9 shows the estimate of the Degree
of Membership (DOM), with the results outlined
below:
DOMx=dm =0.45
DOMx=df =0.25
DOMy=vrs =0.60
DOMy=vrm =0.25
b. Rule Evaluation
This phase aims to validate the intersection
of the inputs within the corresponding rules. The
fuzzy operator employed to obtain the single DOM
representation of the rule is the AND fuzzy
operation, signifying the intersection of fuzzy sets.
The procedure is delineated by the formula shown
in (4).
DOMAUB(x) = min[DOMA(x),DOMB(x)]

(4)

Figure 10 shows the specifics captured by
the Rule Inference tab in MATLAB’s Fuzzy Logic
Designer. The result of the DOM is determined in
the following calculation.
DOMdm∪vrs (Rules2)=min[0.45,0.60]=0.45
DOMdf∪vrs (Rules3)=min[0.25,0.60]=0.25
DOMdm∪vrm (Rules5)=min[0.45,0.25]=0.25
DOMdf∪vrm (Rules6)=min[0.25,0.25]=0.25
c.

Rule Outputs Aggregation
Throughout the aggregation process, the
results of all rules will be combined. The method
will combine the membership functions of any rule
consequents that have been previously modified,
generating the unified fuzzy set ∑ 𝐷𝑂𝑀. Figure 10
illustrates the outcome of the fuzzy set.
According to the aggregate outcome, only
three Membership Functions (MF) cross within the
fuzzy set. These entities are categorized as Safe,
Light, and Moderate. Calculating the parameter's

d. Defuzzification
The defuzzification process utilizes the
aggregated fuzzy output set as input to generate
a specific numerical output. The defuzzification
phase of this calculation applies the centroid
method, as outlined in Equation 3. C(zk)
represents the degree contributions of each
membership zk; hence, the crisp output is
computed as outlined in the subsequent
calculations.

Figure 10. Aggregation of the Outputs
DOM(zA )=DOM(Rules3)=0.25
DOM(zB )=DOM(Rules2)=0.45
DOM(zC )=DOM(Rules5)=0.25
𝑑𝑪𝑨 (C)=DOM(zA )zA + DOM(zB )zB + DOM(zC )zC
/ DOM(z1 )+DOM(z2 )+DOM(z3 )
dCA (C)=35
The result of the manually calculated fuzzy
decisions has become dCA(C) = 35. The manual
calculation result remains consistent once the
system has been finalized. This result indicates
that the MATLAB system operates efficiently in
alignment with basic concepts.
Second Final Step: Optimization and Fine-Tuning
The second final step in the fuzzy
development process is to optimize and refine the
fuzzy system. The technique offers extensive
sampling evaluations across multiple scenarios to
ensure the system operates as intended.
A prior study published in the IEEE journal
suggests that the tuning process for Fuzzy Logic
Controllers lacks a systematic methodology,
necessitating a trial-and-error approach that
involves modifying fuzzy rules and mapping
membership functions until satisfactory outcomes
are achieved.
This technique will be incorporated into the
tuning method of this research. The fine-tuning
process employs a trial-and-error approach to
enhance the fuzzy logic rules and membership
functions until acceptable outcomes are achieved.
a. Rules Optimization
This phase involves assessing a set of
existing rules that specify the system's decisionmaking architecture. This stage primarily aims to

W. I. Susanto et al., Driver assistance collision warning system using a LIDAR sensor …

593

SINERGI Vol. 29, No. 3, October 2025: 587-598

improve the "THEN" statement part of the rules,
which define the brake decision output. The
optimization requires recalibrating the rules to
guarantee that outputs are adjusted to extreme
values. Rules Mapping Optimization is shown in
Figure 11. Figure 12 illustrates the results of rule
optimization.
Three rules, specifically rule2, rule5, and
rule9, had modifications as a result of fine-tuning.
The rules’ output has been adjusted to
accommodate for extreme values. The aim of
producing these extreme values in the rules’
output is to assist with centroid method of
defuzzification, which calculates the mean of the
aggregated outputs, consequently creating an
appropriate brake recommendation.
b. Membership Function Optimization
Following an adjustment of rules, work is
directed towards the optimization of membership
functions. Throughout the tuning implementation
attempt, modifications should prioritize output
adjustment. The accuracy of the system has
limitations, as the centroid method for
defuzzification limits the extraction of the extreme
values (minimum and maximum) required for
specific driving conditions.
To address the matter, the membership
function for the braking decision has been
adjusted by extending the range of these attributes
further the original 0 to 100% range. The
adjustments of the membership function for the
braking choice are illustrated in Figure 12.

The enhanced adjustment simply modifies
the minimum and maximum values, enabling the
system to effectively gather data and respond to
more extreme circumstances with a broader range
of braking applications. This improvement ensures
that the output can be converted into a more
accurate braking action, thereby enhancing the
overall efficiency and precision of the system in
real-world driving situations.
Integrated System:
An integrated system of sensors and
actuators with kinematics and perception
algorithms has been constructed. The final phase
of
the
integration
build
signifies
the
comprehensive integration of the system. The
integration begins with translating the Fuzzy
System developed in MATLAB into Arduino
language, followed by integrating the code to
make it work together.
Figure 13 illustrates the operational logic of
a Driver Assistance Collision Warning System built
in Arduino, which integrates kinematics and
perception algorithms. The primary assessment is
conducted by evaluating the time-to-collision
(TTC). If the TTC is less than 2 seconds, the
system will apply a kinematic algorithm to
determine the result. When the TTC exceeds 2
seconds, the system applies fuzzy-perceptionbased logic to determine the proper output.

Figure 11. Rules Mapping Optimization

Figure 13. Kinematics and Perception Algorithm
Integration System Flowchart
Figure 12. Membership Function Optimization

594

W. I. Susanto et al., Driver assistance collision warning system using a LIDAR sensor …

p-ISSN: 1410-2331 e-ISSN: 2460-1217

Integration Test
Subsequent to the integration build. The
integration testing shall be performed to validate
the seamless connection between the kinematics
algorithm (time-to-collision threshold) and the
perception algorithm (fuzzy logic). This phase
ensures the absence of interference between the
two algorithms.
Test Condition: The collision warning system is
positioned atop the model remote-controlled car
and oriented towards the solid barrier. The
remote-controlled car nears the solid barrier,
resulting in a change in the distance
measurement.
Acceptance Criteria: The system’s warning shall
precisely represent the system's algorithm. The
system's output warning must correspond with its
logic while executing kinematics or perception
decisions. The criteria table for the system is
presented in Table 8.
The system demonstrated its capability for
accurate decision-making in many scenarios, as
indicated by the outcomes shown in Table 9. The
system's integration test passed under the
specified conditions. The system proved to make
correct and precise choices during the testing
scenarios, thereby confirming its adherence to the
established acceptance requirements.

Acceptance Criteria: The system must accurately
calculate relative speed and time-to-collision.
Furthermore, the system's output must adhere to
the system's logic, whether the decision is
kinematics or perception algorithms.
Data Source: The data was obtained from the
integration test, including time-to-collision results
and other essential information.
Fuzzy Output Calculation: signifies utilizing the
rule inference feature of the MATLAB Fuzzy Logic
Designer by inputting the crisp values of distance
and relative speed to obtain the brake decision
output value, as shown in Figure 14.

System Test
The objective of the system test phase is to
validate the precision of distance information
passing from the LIDAR to the Arduino, assess the
accuracy of relative speed and time-to-collision
computations, and confirm that the fuzzy logic
system provides ideal warning responses.

c. Test-3
dA = 215, dB = 363
tA = 18.744 ≈ 19, tB = 19.757 ≈ 20
vr = (363-215)/(20-19) = 148 cm/s
TTC = 363/148 = 2.45 s

Table 8. Integration Test Criteria
TTC
(sec)

Algorithm
Decision

Warning Output

0

Kinematics

Brake – Green

0 < TTC < 1

Kinematics

Retarder – Amber, Buzzer

1 < TTC < 2

Kinematics

Retarder – Amber

TTC > 2

Perception

Brake – Green/Amber/Red

Table 9. Integration Test Result
Test TTC
Fuzzy
No. (cm/s) Output

Decision
Algorithm

Indicator

Judge

1

2.81

55.25

Perception

Brake – Red

Pass

2

0.44

-

Kinematics

Retarder,
Buzzer

Pass

3

2.45

70.00

Perception

Brake – Red

Pass

4

2.53

62.51

Perception

Brake – Red

Pass

Calculation: utilizing basic equations to confirm
the system's computation result. The TTC can be
calculated by applying (2).
a. Test-1
dA = 455, dB = 707
tA = 06.649 ≈ 7, tB = 07.660 ≈ 8
vr = (707-455)/(8-7) = 252 cm/s
TTC = 707/252 = 2.81 s
b. Test-2
dA = 707, dB = 215
tA = 07.660 ≈ 8, tB = 08.689 ≈ 9
vr = (215-707)/(9-8) = -492 cm/s
TTC = |215/-492| = 0.44 s

d. Test-4
dA = 363, dB = 601
tA = 19.757 ≈ 20, tB = 20.760 ≈ 21
vr = (601-363)/(21-20) = 238 cm/s
TTC = 601/238 = 2.53 s
The accuracy of the decision-making
algorithms in the system is validated by comparing
the manual calculations, which employ basic
equations, with the system's computation result.
Table 10 shows that there are no
deviations in Time-to-Collision values between the
system output and the manual calculation.

Figure 14. MATLAB Fuzzy Result Test No. 1, 3, 4

W. I. Susanto et al., Driver assistance collision warning system using a LIDAR sensor …

595

SINERGI Vol. 29, No. 3, October 2025: 587-598

Table 10. System Test Result
System Result
Test
No.

TTC
Fuzzy
(cm/s) Result

Table 11. Acceptance Test Result

Validation
TTC
Manual
(cm/s)

MATLAB
Output

Judge

1

2.81

55.25

2.81

55.3

Pass

2

0.44

-

0.44

-

Pass

3

2.45

70.00

2.45

70

Pass

4

2.53

62.51

2.53

62.5

Pass

Furthermore, the fuzzy output yields the
same results as the system output and MATLAB
calculations, with the exception of test number 2,
where the Time-to-Collision is less than 2
seconds, utilizing a kinematics approach and not
requiring fuzzy output.
The successful test results serve as
concrete proof of the system's capability to
operate with accuracy as well as reliability in
practical conditions, ensuring both safety and
performance. The system test has been assessed
as Pass.
Acceptance Test
An acceptance test verifies that the system
meets end-user requirements and operates
reliably across various environmental conditions.
The collision warning system employs a LIDAR
sensor as its primary component; hence, the
acceptance test will concentrate on LIDAR
performance. According to the research by Park J.
et al. from Hyundai Motor Company, it contains
eight distinct environmental factors that could be
used to evaluate the performance of LIDAR: cover
contamination, strong sunlight, high temperature,
low
temperature,
vibration,
interference,
reflectivity of a target, and transitions between day
and night [26].
To highlight the driver experience, the
acceptance tests for the system will incorporate
real-world driving scenarios with three tests
performance evaluations. The initial assessment
examines cover contamination, with fog, rain, and
dust. The second test features strong sunlight, and
the third test incorporates the transition from day
to night.
In the preliminary phase, conduct tests 1
through 3 to simulate cover contamination by fog,
rain, and dust. The initial test approximated cover
contamination by generating foggy situations. The
container was filled with artificial smoke to
generate a dense fog, significantly reducing
visibility and simulating actual fog conditions. The
second test examines contamination caused by
rain.

596

Test
No.

Req. &
Test

Spec.
Std.
(cm)

Deviation
Result
(cm)

Remark

Judge

1

Fog

±5

±2

No
intervention

Pass

2

Rain

±5

±1

No
intervention

Pass

3

Dust

±5

±5

No
intervention

Pass

4

Strong
Sunlight

±5

±2

No
intervention

Pass

5

Daynight
transition

±5

±3

No
intervention

Pass

A manual droplet generation equipment
was constructed to simulate the process of rainfall
by producing artificial raindrops. The third test
concentrated on dust pollution. In this scenario,
the sensor was obscured by an acrylic cover that
had been previously coated with powder to
simulate a dusty environment. The fourth test
simulated prompt day-night transitions. The
lighting in the chamber executed two prompt
transitions, alternating between brightness and
darkness. This simulation aims to replicate the
illumination change during the transition from day
to night or vice versa. At last, the fifth test
subjected the sensor to strong sunlight conditions.
A flashlight was employed in the same chamber to
illuminate the sensor from multiple angles,
simulating the effects of direct sunlight and glare.
The test results in Table 11 indicate that the
sensor-read value deviation is within ±5 cm,
signifying that the deviation remains within the
LIDAR sensor standard. In conclusion, the sensor
successfully passed the acceptance test
performed under simulated real-world conditions,
indicating excellent reliability and performance
across all assessed scenarios, hence meeting the
acceptance criteria.
Discussion
This research effectively met its goals of
enhancing
vehicle
safety
through
the
establishment of a universally applicable solution.
The main results of the research into the
development of a driver assistance collision
warning system are summarized in the
subsequent items:
The driver assistance collision warning
mechanism, utilising the Arduino Mega 2560 as its
microcontroller, operates on a 5 VDC power
supply, ensuring system interoperability and
adaptability across various vehicle brands and
categories. Its power can be sourced through a

W. I. Susanto et al., Driver assistance collision warning system using a LIDAR sensor …

p-ISSN: 1410-2331 e-ISSN: 2460-1217

vehicle’s electrical port, a USB charging interface,
or an auxiliary power source such as a portable
battery pack. During the calibration phase, the
LIDAR sensor gives highly accurate distance
measurements with a margin of error of ±5 cm,
while the Arduino Microcontroller precisely
executes computational processes.
The integration of the LIDAR sensor’s
precise distance measurement with the Arduino
Mega 2560’s calculations for time-to-collision
estimation, relative speed, and fuzzy logic output
facilitates reliable data for kinematics and
perception algorithms. The kinematics and
perception
algorithms
were
effectively
harmonized, as verified through comprehensive
integration testing, which confirmed their
seamless
collaboration.
The
successful
completion of these evaluations underscores the
system's dependability and the strong cohesion of
its fundamental components.
The
collision
warning
mechanism
demonstrates its precision through systematic
validation, wherein its outputs are benchmarked
against manual computations, ensuring consistent
and reliable decision-making for both kinematic
modeling and perceptual algorithms. The system's
accuracy was rigorously evaluated in simulated
actual environments using the acceptance test.
The sensors precisely identified objects under
various circumstances, including fog, rain, dust,
and contrasting brightness levels during both day
and night, as well as bright sunlight. This
exceptional performance across varied settings
highlights the system's resistance and reliability.
The system's primary objective is to provide users
with early warnings to avoid collisions between
vehicles. By precisely identifying potential
collisions in advance, the system allows drivers to
implement crucial precautions to prevent
accidents, hence improving overall road safety.
CONCLUSION
In conclusion, the system demonstrates a
significant enhancement in the technology used to
improve vehicle safety. With its broad compatibility
across different vehicle brands and segments, its
seamless
integration
of
decision-making
algorithms, and its high-performance accuracy in
real-world applications, it serves as an essential
instrument for enhancing vehicle safety. By
integrating kinematics and perception, this study
overcomes the limitations of previous research,
achieving real-time accuracy while reducing
reliance on pre-trained models. This research
achieves its primary objectives and significantly

advances the broader goal of enhancing road
safety for all vehicle drivers.
ACKNOWLEDGMENT
I am deeply thankful to the Swiss German
University for the curriculum and instruction
program, which has offered me a stimulating and
enriching academic experience. My gratitude
extends to Polytron for providing me with the
practical insights and professional environment
that have enriched my research experience.
REFERENCES
[1] Y. Lobanova and S. Evtiukov, “Role and
methods of accident ability diagnosis in
ensuring traffic safety,” Transportation
Research Procedia, vol. 50, pp. 363–372,
2020, doi: 10.1016/j.trpro.2020.10.043.
[2] W. I. Susanto, H. Nasution, and T. D. Sofianti,
“Driver Assistance Collision Warning System
Using LIDAR Sensor with Kinematic and
Perception Algorithms,” Thesis, Swiss
German University, 2024.
[3] I. Isaksson Hellman and M. Lindman,
“Estimating the crash reducing effect of
Advanced Driver Assistance Systems
(ADAS) for vulnerable road users,” Traffic
Safety Research, vol. 4, p. 000036, Nov.
2023, doi: 10.55329/blzz2682.
[4] H. Kunto, H. Nasution, and D. Hendriana,
“Design and Develop Vehicle Anti-Collision
Warning
System,”
Conference
on
Management and Engineering in Industry
(CMEI), vol. 4, no. 1, pp. 19–24, Aug. 2022.
[5] Y. Yuan, Y. Lu, and Q. Wang, “Adaptive
forward vehicle collision warning based on
driving behavior,” Neurocomputing, vol. 408,
pp.
64–71,
Sep.
2020,
doi:
10.1016/j.neucom.2019.11.024.
[6] G. Šimić, “Can AI be conscious,” Abstract
book of A4L Symposium in Zagreb 2024,
2024.
[7] L. Masello, G. Castignani, B. Sheehan, F.
Murphy, and K. McDonnell, “On the road
safety benefits of advanced driver assistance
systems in different driving contexts,” Transp
Res Interdiscip Perspect, vol. 15, p. 100670,
Sep. 2022, doi: 10.1016/j.trip.2022.100670.
[8] C. Salazar Dorn and C. L. Knüvener,
“Building the Bridge Between Automotive
SPICE® and Agile Development,” Systems,
Software
and
Services
Process
Improvement, 2021, pp. 391–405. doi:
10.1007/978-3-030-85521-5_26.
[9] L. J. Moukahal, M. A. Elsayed, and M.
Zulkernine, “Vehicle Software Engineering

W. I. Susanto et al., Driver assistance collision warning system using a LIDAR sensor …

597

SINERGI Vol. 29, No. 3, October 2025: 587-598

(VSE): Research and Practice,” IEEE Internet
Things Journal, vol. 7, no. 10, pp. 10137–
10149,
Oct.
2020,
doi:
10.1109/JIOT.2020.3001026.
[10] S. Das and A. K. Maurya, “Defining Time-toCollision Thresholds by the Type of Lead
Vehicle
in
Non-Lane-Based
Traffic
Environments,” IEEE Transactions on
Intelligent Transportation Systems, vol. 21,
no. 12, pp. 4972–4982, Dec. 2020, doi:
10.1109/TITS.2019.2946001.
[11] J. Cordes, M. Chraibi, A. Tordeux, and A.
Schadschneider, “Time-To-Collision Models
for Single-File Pedestrian Motion,” Collect
Dyn, vol. 6, p. A133, Jan. 2022, doi:
10.17815/CD.2021.133.
[12] United Nations. Economic Commission for
Europe, World Forum for Harmonization of
Vehicle Regulations (WP. 29) : how it works,
how to join it., 4th ed. 2022.
[13] United Nations. Economic Commission for
Europe, Uniform provisions concerning the
approval of audible warning devices and of
motor vehicles with regard to their audible
signals. 2020. Accessed: Jul. 17, 2024.
[Online].
Available:
https://unece.org/transport/documents/2021/
05/standards/un-regulation-no-28-amend6
[14] United Nations. Economic Commission for
Europe, Uniform provisions concerning the
approval of vehicles with regard to the
location and identification of hand controls,
tell-tales and indicators. 2023. Accessed: Jul.
17,
2024.
[Online].
Available:
https://unece.org/transport/documents/2023/
12/standards/addendum-120-un-regulationno-121-revision-2-amendment-6
[15] J. F. McDonald, “Following Rules and Traffic
Flow: What If Everyone Used the ThreeSecond Rule?,” SSRN Electronic Journal,
2023, doi: 10.2139/ssrn.4540900.
[16] K. YOSHIOKA, “A Tutorial and Review of
Automobile Direct ToF LiDAR SoCs:
Evolution of Next-Generation LiDARs,” IEICE
Transactions on Electronics, vol. E105.C, no.
10, p. 2021CTI0002, Oct. 2022, doi:
10.1587/transele.2021CTI0002.
[17] Z. Dai, A. Wolf, P.-P. Ley, T. Glück, M. C.
Sundermeier,
and
R.
Lachmayer,
“Requirements for Automotive LiDAR
Systems,” Sensors, vol. 22, no. 19, p. 7532,
Oct. 2022, doi: 10.3390/s22197532.

598

[18] H. K. Kondaveeti, N. K. Kumaravelu, S. D.
Vanambathina, S. E. Mathe, and S.
Vappangi, “A systematic literature review on
prototyping with Arduino: Applications,
challenges, advantages, and limitations,”
Comput Sci Rev, vol. 40, p. 100364, May
2021, doi: 10.1016/j.cosrev.2021.100364.
[19] M. Markofsky, M. Schäfer, and D. Schramm,
“Use Cases and Methods of Virtual
ADAS/ADS Calibration in Simulation,”
Vehicles, vol. 5, no. 3, pp. 802–829, Jul.
2023, doi: 10.3390/vehicles5030044.
[20] A. Goeritno and I. Setyawibawa, “An
Electronic Device Reviewed by Diagnosing
on the Modules Embodiment,” International
Journal of Electronics and Communications
Systems, vol. 1, no. 2, pp. 41–55, Dec. 2021,
doi: 10.24042/ijecs.v1i2.10383.
[21] D. Romahadi, D. Feriyanto, F. Anggara, F. P.
Wijaya, and W. Dong, “Intelligent system
design for identification of unbalance and
misalignment using Fuzzy Logic methods,”
SINERGI, vol. 28, no. 2, p. 241, Apr. 2024,
doi: 10.22441/sinergi.2024.2.004.
[22] H. Nasution, H. Jamaluddin, and J. Mohd.
Syeriff, “Energy Analysis for Air Conditioning
System Using Fuzzy Logic Controller,”
TELKOMNIKA
(Telecommunication
Computing Electronics and Control), vol. 9,
no. 1, p. 139, Apr. 2011,
doi:
10.12928/telkomnika.v9i1.680.
[23] V. Mateichyk, N. Kostian, M. Smieszek, J.
Mosciszewski,
and
L.
Tarandushka,
“Evaluating Vehicle Energy Efficiency in
Urban Transport Systems Based on Fuzzy
Logic Models,” Energies (Basel), vol. 16, no.
2,
p.
734,
Jan.
2023,
doi:
10.3390/en16020734.
[24] J. J. Cardiel-Ortega and R. Baeza-Serrato,
“Failure Mode and Effect Analysis with a
Fuzzy Logic Approach,” Systems, vol. 11, no.
7,
p.
348,
Jul.
2023,
doi:
10.3390/systems11070348.
[25] G. J. Klir and B. Yuan, “Fuzzy sets and fuzzy
logic - theory and applications,” Prentice Hall,
1995.
[Online].
Available:
https://api.semanticscholar.org/CorpusID:46
622061
[26] J. Park, J. Cho, S. Lee, S. Bak, and Y. Kim,
“An Automotive LiDAR Performance Test
Method in Dynamic Driving Conditions,”
Sensors, vol. 23, no. 8, p. 3892, Apr. 2023,
doi: 10.3390/s23083892.

W. I. Susanto et al., Driver assistance collision warning system using a LIDAR sensor …