Content uploaded by Paolo Visconti
Author content
All content in this area was uploaded by Paolo Visconti on Jan 22, 2025
Content may be subject to copyright.
Available via license: CC BY 4.0
Content may be subject to copyright.
Academic Editor: Damianos Gavalas
Received: 17 December 2024
Revised: 11 January 2025
Accepted: 16 January 2025
Published: 19 January 2025
Citation: Visconti, P.; Rausa, G.;
Del-Valle-Soto, C.; Velázquez, R.;
Cafagna, D.; De Fazio, R. Innovative
Driver Monitoring Systems and
On-Board-Vehicle Devices in a
Smart-Road Scenario Based on the
Internet of Vehicle Paradigm: A
Literature and Commercial Solutions
Overview. Sensors 2025,25, 562.
https://doi.org/10.3390/s25020562
Copyright: © 2025 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license
(https://creativecommons.org/
licenses/by/4.0/).
Review
Innovative Driver Monitoring Systems and On-Board-Vehicle
Devices in a Smart-Road Scenario Based on the Internet of
Vehicle Paradigm: A Literature and Commercial
Solutions Overview
Paolo Visconti 1,* , Giuseppe Rausa 1, Carolina Del-Valle-Soto 2, Ramiro Velázquez 3, Donato Cafagna 1
and Roberto De Fazio 1,3
1Department of Innovation Engineering, University of Salento, 73100 Lecce, Italy;
giuseppe.rausa@unisalento.it (G.R.); donato.cafagna@unisalento.it (D.C.);
roberto.defazio@unisalento.it (R.D.F.)
2Facultad de Ingeniería, Universidad Panamericana, Zapopan 45010, Mexico; cvalle@up.edu.mx
3Facultad de Ingeniería, Universidad Panamericana, Aguascalientes 20296, Mexico; rvelazquez@up.edu.mx
*Correspondence: paolo.visconti@unisalento.it; Tel.: +39-0832-29-7334
Abstract: In recent years, the growing number of vehicles on the road have exacerbated
issues related to safety and traffic congestion. However, the advent of the Internet of
Vehicles (IoV) holds the potential to transform mobility, enhance traffic management and
safety, and create smarter, more interconnected road networks. This paper addresses key
road safety concerns, focusing on driver condition detection, vehicle monitoring, and
traffic and road management. Specifically, various models proposed in the literature for
monitoring the driver’s health and detecting anomalies, drowsiness, and impairment due
to alcohol consumption are illustrated. The paper describes vehicle condition monitoring
architectures, including diagnostic solutions for identifying anomalies, malfunctions, and
instability while driving on slippery or wet roads. It also covers systems for classifying
driving style, as well as tire and emissions monitoring. Moreover, the paper provides a
detailed overview of the proposed traffic monitoring and management solutions, along
with systems for monitoring road and environmental conditions, including the sensors
used and the Machine Learning (ML) algorithms implemented. Finally, this review also
presents an overview of innovative commercial solutions, illustrating advanced devices for
driver monitoring, vehicle condition assessment, and traffic and road management.
Keywords: internet of vehicles; intelligent transport system; wearable device; driver
monitoring; traffic management; OBD-II; vehicle condition monitoring
1. Introduction
In recent years, growing urbanization and the increasing number of vehicles on the
road have introduced new challenges related to transportation safety, efficiency, and sus-
tainability, leading to growing interest in the Internet of Vehicles (IoV) [
1
]. Within this
framework, the monitoring systems of driver, vehicle, and road conditions have become
key components of Intelligent Transportation Systems (ITSs) and smart cities [
2
]. These
systems leverage advanced technologies, including sensors, IoT devices, and Artificial
Intelligence (AI) to collect, analyze, and utilize real-time data. Their objectives include
optimizing traffic flows, enhancing road safety, safeguarding drivers, and reducing envi-
ronmental impact (Figure 1) [
3
]. The automotive industry has recently undergone rapid
Sensors 2025,25, 562 https://doi.org/10.3390/s25020562
Sensors 2025,25, 562 2 of 57
advancements in driver assistance technologies, notably with the development of advanced
driver assistance systems (ADASs) [
4
,
5
]. These systems represent a crucial step toward fu-
ture mobility [
6
]. However, addressing road safety, transportation efficiency, sustainability,
and increasingly complex driving conditions requires innovative solutions, particularly to
ensure the safety of all road users [
7
]. In this regard, Driver Monitoring Systems (DMSs) are
gaining prominence for their ability to detect health anomalies, drowsiness, intoxication,
and unsafe driving behaviors [
8
]. Recent studies indicate that human errors account for
approximately 90% of road accidents, many of which could be prevented with effective
monitoring systems [
9
,
10
]. These systems rely on both intrusive and non-intrusive sensors
to monitor drivers’ biophysical parameters (such as Electrocardiograms (ECGs), photo-
plethysmography (PPG), Electrodermal Activity (EDA), heart rate (HR), Blood Alcohol
Concentration (BAC), respiratory rate (RR), and oxygen saturation (SpO
2
)) and cameras to
detect facial cues like yawning, blink rate, and head movements. Using Machine Learning
(ML) algorithms and real-time data analysis, these systems enable timely interventions,
preventing accidents and saving lives [
11
–
15
]. Research has also addressed the recurring
tragedy of infants being accidentally left in vehicles. Innovative systems now exist that
can detect an infant’s cries and send alert signals through mobile applications, prompting
immediate action [
16
,
17
]. Investing in such technologies enhances road safety and fosters a
more responsible and conscientious driving environment.
Sensors 2025, 25, x FOR PEER REVIEW 2 of 62
optimizing traffic flows, enhancing road safety, safeguarding drivers, and reducing envi-
ronmental impact (Figure 1) [3]. The automotive industry has recently undergone rapid
advancements in driver assistance technologies, notably with the development of ad-
vanced driver assistance systems (ADASs) [4,5]. These systems represent a crucial step
toward future mobility [6]. However, addressing road safety, transportation efficiency,
sustainability, and increasingly complex driving conditions requires innovative solutions,
particularly to ensure the safety of all road users [7]. In this regard, Driver Monitoring
Systems (DMSs) are gaining prominence for their ability to detect health anomalies,
drowsiness, intoxication, and unsafe driving behaviors [8]. Recent studies indicate that
human errors account for approximately 90% of road accidents, many of which could be
prevented with effective monitoring systems [9,10]. These systems rely on both intrusive
and non-intrusive sensors to monitor drivers’ biophysical parameters (such as Electrocar-
diograms (ECGs), photoplethysmography (PPG), Electrodermal Activity (EDA), heart
rate (HR), Blood Alcohol Concentration (BAC), respiratory rate (RR), and oxygen satura-
tion (SpO
2
)) and cameras to detect facial cues like yawning, blink rate, and head move-
ments. Using Machine Learning (ML) algorithms and real-time data analysis, these sys-
tems enable timely interventions, preventing accidents and saving lives [11–15]. Research
has also addressed the recurring tragedy of infants being accidentally left in vehicles. In-
novative systems now exist that can detect an infant’s cries and send alert signals through
mobile applications, prompting immediate action [16,17]. Investing in such technologies
enhances road safety and fosters a more responsible and conscientious driving environ-
ment.
Figure 1. Summary picture of the IoV paradigm in the smart city scenario for road safety and trans-
portation efficiency purposes: driver, vehicle, road, and traffic monitoring systems.
In this context, advanced on-board diagnostic (OBD-II) technologies play a pivotal
role, not only enhancing vehicle reliability and durability but also directly improving road
user safety [18,19]. Many innovative systems utilize data collected through the vehicle’s
OBD-II system to monitor its status. Another significant aspect of these technologies is
emissions monitoring, which is increasingly important given the tightening of environ-
mental regulations worldwide. Real-time detection of polluting emissions helps minimize
environmental impact and optimize resource use [20]. Tire pressure and wear monitoring
Figure 1. Summary picture of the IoV paradigm in the smart city scenario for road safety and
transportation efficiency purposes: driver, vehicle, road, and traffic monitoring systems.
In this context, advanced on-board diagnostic (OBD-II) technologies play a pivotal
role, not only enhancing vehicle reliability and durability but also directly improving road
user safety [
18
,
19
]. Many innovative systems utilize data collected through the vehicle’s
OBD-II system to monitor its status. Another significant aspect of these technologies is
emissions monitoring, which is increasingly important given the tightening of environ-
mental regulations worldwide. Real-time detection of polluting emissions helps minimize
environmental impact and optimize resource use [
20
]. Tire pressure and wear monitoring
extend beyond routine maintenance. Proper tire management improves fuel efficiency and
reduces the risk of accidents caused by blowouts or loss of control [
21
]. Similarly, vehicle
Sensors 2025,25, 562 3 of 57
stability is a fundamental safety factor. Real-time monitoring systems can detect anomalies
in driving parameters, preventing potentially hazardous situations [22].
Traffic and road condition monitoring are essential for ensuring urban mobility’s
safety, efficiency, and sustainability. Advanced systems incorporating sensors, cameras,
and real-time communication technologies collect and analyze data on traffic patterns
and infrastructure conditions [
23
]. This information exchange has given rise to interactive
Vehicle-to-Infrastructure (V2I) communication networks, leveraging protocols like Wi-Fi,
LoRa (Long Range), 4G/5G, and other technologies to support seamless connectivity and
enhanced mobility [
24
]. Another critical aspect is traffic management, which focuses on op-
timizing vehicle flow, reducing congestion, and enhancing the driving experience through
intelligent planning and innovative solutions such as smart traffic lights and dynamic
signaling systems [
25
]. Additionally, vehicular traffic has been harnessed as a source of
energy, enabling the development of systems that harvest, store, and supply energy to
power road infrastructure [
26
]. Monitoring road conditions is equally vital, allowing for the
timely detection and resolution of issues like potholes and asphalt deterioration, thereby
improving both safety and comfort for road users [
27
]. Implemented cooperatively, all pre-
vious practices enhance urban living standards and support environmental sustainability
by fostering smarter, more responsible mobility solutions [28].
Vehicle health monitoring systems are an essential aid to the driver since they provide
safety support, such as DMSs used to detect driver anomalies, like anxiety, drowsiness,
fatigue, or drunkenness, which are all conditions that significantly reduce attention while
driving. DMSs combined with a traffic monitoring and management system create smart
frameworks for improving road safety, as by monitoring the vehicles’ flow on the roads
they can provide alternative routes in case of emergency, reducing the risk of accidents.
Also, the driver’s condition information can be jointly processed and evaluated along with
the vehicle parameters to implement smarter DMSs, determining a cumulative driver risk
that also takes into account the vehicle conditions (brake wear, tire pressure and wear, etc.).
For these reasons, the proposed manuscript explores recent advancements in intelligent
driver, vehicle, and traffic monitoring systems to bring out insights and trends for the
development of road safety systems in the near future.
This manuscript aims to provide the reader with an overview of IoV solutions in the
ITS and smart cities paradigm, analyzing three different macro categories:
➢
Driver monitoring: various on-board applications for driver monitoring are presented,
focusing on detecting health anomalies, drowsiness, stress, and impairment due to
alcohol consumption. Additionally, the intrusiveness of these applications and their
impact on user ergonomics are analyzed (Section 2).
➢
Vehicle condition monitoring: on-board technical solutions for vehicle condition
monitoring (i.e., tire pressure and wear, exhaust emissions, and vehicle stability) are
examined alongside innovative diagnostic systems and devices for detecting road
conditions. Additionally, systems for monitoring driving style and driver behavior
are analyzed (Section 3).
➢
Road and environmental monitoring, traffic management: an overview is presented
of innovative off-board systems for traffic management, environmental monitoring,
and road monitoring, including predictive maintenance solutions. These systems also
offer capabilities for detecting road congestion and providing drivers with alternative
routes (Section 4).
To enhance the manuscript’s scientific contribution, each section includes a compara-
tive analysis of the reviewed articles, along with summary tables highlighting their key
features. Additionally, Section 5presents innovative commercial solutions developed by
international companies for monitoring drivers, vehicles, and traffic conditions.
Sensors 2025,25, 562 4 of 57
The structure of this review paper is as follows. Section 1introduces the topics covered
and explains the methodology used for selecting the scientific articles. Section 2provides
an overview of on-board solutions for driver monitoring, organized into three subsections
focusing on applications for detecting driver status. Section 3examines on-board solutions
for vehicle condition monitoring, divided into five subsections covering vehicle diagnostics,
asphalt monitoring systems, driving style and behavior analysis, tire monitoring, and
emissions monitoring. Section 4explores solutions for traffic monitoring and management,
as well as road and environmental condition monitoring. Section 5reviews innovative com-
mercial applications related to the discussed topics. Finally, Section 6presents comments
and conclusions, emphasizing potential future applications.
Selection Method of Analyzed Articles Based on PRISMA Methodologies
This section outlines the criteria used for selecting scientific articles, focusing on key
aspects such as relevance to the covered topics, publication year, and redundancy with
other works. The objective of this review is to provide readers with a comprehensive, state-
of-the-art overview of IoV and ITS paradigm applications. The article selection process
followed the PRISMA methodology, ensuring the reliability and applicability of the chosen
approach [
29
,
30
]. The selection process began with evaluating the title of each candidate
article to identify keywords related to the topic. Next, the abstract was reviewed to assess
its alignment with the issues addressed in this review. If the article appeared relevant, it was
thoroughly read and analyzed. Additional sources were consulted to resolve ambiguities
for articles with unclear topics. If the content remained unclear after further investigation,
the article was excluded.
Figure 2a illustrates the selection process, which included assessing the title’s relevance,
the abstract’s affinity, and the manuscript’s scientific value. Figure 2b depicts the main
keywords used to filter the literature. This methodology was consistently applied across all
topics discussed in this review.
Sensors 2025, 25, x FOR PEER REVIEW 4 of 62
features. Additionally, Section 5 presents innovative commercial solutions developed by
international companies for monitoring drivers, vehicles, and traffic conditions.
The structure of this review paper is as follows. Section 1 introduces the topics cov-
ered and explains the methodology used for selecting the scientific articles. Section 2 pro-
vides an overview of on-board solutions for driver monitoring, organized into three sub-
sections focusing on applications for detecting driver status. Section 3 examines on-board
solutions for vehicle condition monitoring, divided into five subsections covering vehicle
diagnostics, asphalt monitoring systems, driving style and behavior analysis, tire moni-
toring, and emissions monitoring. Section 4 explores solutions for traffic monitoring and
management, as well as road and environmental condition monitoring. Section 5 reviews
innovative commercial applications related to the discussed topics. Finally, Section 6 pre-
sents comments and conclusions, emphasizing potential future applications.
Selection Method of Analyzed Articles Based on PRISMA Methodologies
This section outlines the criteria used for selecting scientific articles, focusing on key
aspects such as relevance to the covered topics, publication year, and redundancy with
other works. The objective of this review is to provide readers with a comprehensive,
state-of-the-art overview of IoV and ITS paradigm applications. The article selection pro-
cess followed the PRISMA methodology, ensuring the reliability and applicability of the
chosen approach [29,30]. The selection process began with evaluating the title of each can-
didate article to identify keywords related to the topic. Next, the abstract was reviewed to
assess its alignment with the issues addressed in this review. If the article appeared rele-
vant, it was thoroughly read and analyzed. Additional sources were consulted to resolve
ambiguities for articles with unclear topics. If the content remained unclear after further
investigation, the article was excluded.
Figure 2a illustrates the selection process, which included assessing the title’s rele-
vance, the abstract’s affinity, and the manuscript’s scientific value. Figure 2b depicts the
main keywords used to filter the literature. This methodology was consistently applied
across all topics discussed in this review.
(a)
Figure 2. Cont.
Sensors 2025,25, 562 5 of 57
Sensors 2025, 25, x FOR PEER REVIEW 5 of 62
(b)
Figure 2. Document selection method: (a) description of articles’ selecting method with topics re-
lated to the presented review paper, (b) main keywords to filter the documents found in the litera-
ture.
To provide an accurate analysis of the topics covered in this manuscript (i.e., on-
board systems for driver and vehicle monitoring, and off-board systems for traffic and
road management and monitoring), the authors have reviewed 119 documents, including
research articles, conference papers, review articles, and websites. The primary sources
for these articles were Elsevier, MDPI, and IEEE. In Figure 3a, the selected articles are
categorized by publisher, while in Figure 3b, they are sorted by type.
(a)
Figure 2. Document selection method: (a) description of articles’ selecting method with topics related
to the presented review paper, (b) main keywords to filter the documents found in the literature.
To provide an accurate analysis of the topics covered in this manuscript (i.e., on-
board systems for driver and vehicle monitoring, and off-board systems for traffic and
road management and monitoring), the authors have reviewed 119 documents, including
research articles, conference papers, review articles, and websites. The primary sources
for these articles were Elsevier, MDPI, and IEEE. In Figure 3a, the selected articles are
categorized by publisher, while in Figure 3b, they are sorted by type.
Sensors 2025, 25, x FOR PEER REVIEW 5 of 62
(b)
Figure 2. Document selection method: (a) description of articles’ selecting method with topics re-
lated to the presented review paper, (b) main keywords to filter the documents found in the litera-
ture.
To provide an accurate analysis of the topics covered in this manuscript (i.e., on-
board systems for driver and vehicle monitoring, and off-board systems for traffic and
road management and monitoring), the authors have reviewed 119 documents, including
research articles, conference papers, review articles, and websites. The primary sources
for these articles were Elsevier, MDPI, and IEEE. In Figure 3a, the selected articles are
categorized by publisher, while in Figure 3b, they are sorted by type.
(a)
Sensors 2025, 25, x FOR PEER REVIEW 6 of 62
(b)
Figure 3. (a) Selected articles sorted by publishers, (b) selected articles sorted by typology.
2. On-Board Driver Monitoring Systems for Road Safety Applications
In recent years, road safety has become increasingly important, driven by the need
to reduce the number of accidents. In this context, on-board DMSs are emerging as a
promising solution to improve the safety of vehicles and occupants [31]. Different meth-
ods have been proposed in the literature to detect the driver’s condition, starting from the
acquisition of biophysical signals or video/images, allowing the detection of drowsiness,
stress, tiredness, and distraction of the driver, i.e., all aspects that reduce the aention to
driving and considerably increase the risk of accidents [32]. The primary goal of DMSs is
to identify potentially hazardous driving behaviors and provide immediate alerts to the
driver; they can also record and analyze the driving data to prevent accidents, promoting
a culture of responsible road safety. Figure 4 illustrates key devices used in DMSs, includ-
ing wearable and non-wearable sensors for collecting biophysical data, and cameras to
capture facial features, head movements, and eye behavior, such as blink rates and eyelid
drooping. ML algorithms process these data to determine stress, drowsiness, alcohol ef-
fects, and the driver’s health condition. A further feature of DMSs is the capability to re-
port hazardous conditions to the driver, authorities, and emergency services. Below, the
methodologies proposed in the literature are analyzed: the intrusive methods involve
wearable devices, while non-intrusive ones rely on image processing and biosignal acqui-
sition by sensors integrated into the seatbelt, steering wheel, and seat.
Monitoring the driver’s condition is crucial for road safety, enabling early detection
of driver anomalies, like drowsiness or drunkenness. These systems, integrated into mod-
ern vehicles, not only improve personal safety but also complement the vehicle’s ad-
vanced diagnostic systems, which detect malfunctions and analyze driving style to pre-
vent risky situations. In addition, systems for monitoring vehicle stability play an essential
role in ensuring road safety, guaranteeing constant control over the vehicle and driving
conditions. These technologies, working in synergy, also contribute to a more efficient
traffic management, optimizing urban mobility and reducing the risk of accidents thanks
to the collection and analysis of data on driving behavior and vehicle conditions.
Review papers:
2024: 3
2023: 8
2022: 2
2019: 1
2018: 1
Conference papers:
2024: 5
2023: 4
2022: 2
2021: 1
2020: 1
2018: 1
2015: 1
Research works:
2024: 19
2023: 23
2022: 15
2021: 7
2020: 7
2019: 2
2018: 1
2013:2
Figure 3. (a) Selected articles sorted by publishers, (b) selected articles sorted by typology.
Sensors 2025,25, 562 6 of 57
2. On-Board Driver Monitoring Systems for Road Safety Applications
In recent years, road safety has become increasingly important, driven by the need
to reduce the number of accidents. In this context, on-board DMSs are emerging as a
promising solution to improve the safety of vehicles and occupants [
31
]. Different methods
have been proposed in the literature to detect the driver’s condition, starting from the
acquisition of biophysical signals or video/images, allowing the detection of drowsiness,
stress, tiredness, and distraction of the driver, i.e., all aspects that reduce the attention to
driving and considerably increase the risk of accidents [
32
]. The primary goal of DMSs is
to identify potentially hazardous driving behaviors and provide immediate alerts to the
driver; they can also record and analyze the driving data to prevent accidents, promoting a
culture of responsible road safety. Figure 4illustrates key devices used in DMSs, including
wearable and non-wearable sensors for collecting biophysical data, and cameras to capture
facial features, head movements, and eye behavior, such as blink rates and eyelid drooping.
ML algorithms process these data to determine stress, drowsiness, alcohol effects, and the
driver’s health condition. A further feature of DMSs is the capability to report hazardous
conditions to the driver, authorities, and emergency services. Below, the methodologies
proposed in the literature are analyzed: the intrusive methods involve wearable devices,
while non-intrusive ones rely on image processing and biosignal acquisition by sensors
integrated into the seatbelt, steering wheel, and seat.
Sensors 2025, 25, x FOR PEER REVIEW 7 of 62
Figure 4. Example of an innovative DMS based on intrusive and non-intrusive methods for the ac-
quisition and processing of biophysical and behavioral driver parameters by using sensors and cam-
eras integrated into the cockpit and wearable devices.
This section explores innovative DMSs highlighted in recent literature. It is organized
into four subsections: driver health monitoring, driver drowsiness monitoring, driver
drunkenness monitoring, and comparative analysis of the reported articles.
2.1. Driver Health Monitoring Systems to Detect Stress, Anxiety, and Other
Biophysical Parameters
This subsection analyzes the applications developed by the authors for monitoring
driver health during driving operations. In this regard, Kaur et al. proposed an IoV-Health
system dedicated to monitoring the health conditions of the driver [11]. The monitoring
occurs with six wearable sensors, such as an Electrocardiogram (ECG) and Blood Pressure
(BP), Body Temperature (BT), heart rate (HR), Respiration Rate (RR), and alcohol sensors.
Based on the data acquired by the sensors, the system can manage four emergency states:
heart failure, stress, drowsiness, and drunkenness. Additionally, the system designed by
the authors issues alerts to nearby vehicles, hospitals, and ambulances according to the
detected alarm level, ensuring prompt emergency assistance for the driver. The authors
used different ML algorithms for the classification of the recorded data, obtaining high
percentages of accuracy through Random Forest (RF) (99.86%), ensemble classifier (EC)
(99.83%), and decision tree (DT) (99.7%) algorithms. Finally, the authors observed that
their system’s efficiency surpassed similar systems by more than 50%.
Affanni et al. presented a smart necklace for monitoring the driver’s well-being [12],
including a sensor that monitors HR and blood oxygen saturation (SpO
2
). The collected
data are sent to a mobile application via Wi-Fi for continuous real-time viewing. The au-
thors focused on the necklace’s comfort, obtaining excellent results from the driving tests
they carried out. The algorithm for the data processing is based on the “Mountaineer al-
gorithm” for pick detection in ECG signals [33]. Gong et al. in [13] adopted remote photo-
plethysmography (rPPG) based on a monochrome camera to measure HR. They proposed
an innovative approach called “Quality-guided Spectrum Peak Screening” (QSPS) to elim-
inate noise interference related to rapidly changing lighting conditions, head movements,
or vibrations of the vehicle. The QSPS framework detects PPG signals on multiple regions
of the face, in order to aenuate the residual noise through a spectral peak screening
Figure 4. Example of an innovative DMS based on intrusive and non-intrusive methods for the
acquisition and processing of biophysical and behavioral driver parameters by using sensors and
cameras integrated into the cockpit and wearable devices.
Monitoring the driver’s condition is crucial for road safety, enabling early detection
of driver anomalies, like drowsiness or drunkenness. These systems, integrated into
modern vehicles, not only improve personal safety but also complement the vehicle’s
advanced diagnostic systems, which detect malfunctions and analyze driving style to
prevent risky situations. In addition, systems for monitoring vehicle stability play an
essential role in ensuring road safety, guaranteeing constant control over the vehicle and
driving conditions. These technologies, working in synergy, also contribute to a more
efficient traffic management, optimizing urban mobility and reducing the risk of accidents
thanks to the collection and analysis of data on driving behavior and vehicle conditions.
Sensors 2025,25, 562 7 of 57
This section explores innovative DMSs highlighted in recent literature. It is organized
into four subsections: driver health monitoring, driver drowsiness monitoring, driver
drunkenness monitoring, and comparative analysis of the reported articles.
2.1. Driver Health Monitoring Systems to Detect Stress, Anxiety, and Other
Biophysical Parameters
This subsection analyzes the applications developed by the authors for monitoring
driver health during driving operations. In this regard, Kaur et al. proposed an IoV-Health
system dedicated to monitoring the health conditions of the driver [
11
]. The monitoring
occurs with six wearable sensors, such as an Electrocardiogram (ECG) and Blood Pressure
(BP), Body Temperature (BT), heart rate (HR), Respiration Rate (RR), and alcohol sensors.
Based on the data acquired by the sensors, the system can manage four emergency states:
heart failure, stress, drowsiness, and drunkenness. Additionally, the system designed by
the authors issues alerts to nearby vehicles, hospitals, and ambulances according to the
detected alarm level, ensuring prompt emergency assistance for the driver. The authors
used different ML algorithms for the classification of the recorded data, obtaining high
percentages of accuracy through Random Forest (RF) (99.86%), ensemble classifier (EC)
(99.83%), and decision tree (DT) (99.7%) algorithms. Finally, the authors observed that their
system’s efficiency surpassed similar systems by more than 50%.
Affanni et al. presented a smart necklace for monitoring the driver’s well-being [
12
],
including a sensor that monitors HR and blood oxygen saturation (SpO
2
). The collected
data are sent to a mobile application via Wi-Fi for continuous real-time viewing. The
authors focused on the necklace’s comfort, obtaining excellent results from the driving
tests they carried out. The algorithm for the data processing is based on the “Mountaineer
algorithm” for pick detection in ECG signals [
33
]. Gong et al. in [
13
] adopted remote
photoplethysmography (rPPG) based on a monochrome camera to measure HR. They pro-
posed an innovative approach called “Quality-guided Spectrum Peak Screening” (QSPS) to
eliminate noise interference related to rapidly changing lighting conditions, head move-
ments, or vibrations of the vehicle. The QSPS framework detects PPG signals on multiple
regions of the face, in order to attenuate the residual noise through a spectral peak screening
algorithm. With this innovative approach, the authors demonstrated that QSPS has greater
robustness than infrared methods, achieving a Mean Absolute Error (MAE) of 4.32 bpm, a
Root Mean Square Error (RMSE) of 6.15 bpm, and a Pearson correlation coefficient of 0.943
in the nighttime test and 0.906 in the daytime test for 10 min time monitoring.
Jiao et al. conducted an experimental study for real-time detection of driver fatigue
status to improve driving and traffic safety [
34
]. The authors conducted an experimental
study to gather data, including fatigue levels measured using the Karolinska Sleepiness
Scale [
35
], as well as Heart Rate Variability (HRV) and Electrodermal Activity (EDA)
parameters. Broad statistical analysis revealed significant variations in several HRV and
EDA features across different fatigue levels. Using various ML techniques, the Light
Gradient Boosting Machine (LGBM) classifier demonstrated the best performance for
binary classification, achieving an accuracy of 88.7% with HRV and EDA features as inputs.
For three-class classification, accuracy slightly dropped to 85.6% with the RF classifier.
These findings highlight the effectiveness of combining HRV and EDA features to capture
diverse physiological responses to fatigue, enhancing overall detection accuracy.
Leicht et al. in [
36
] evaluated non-intrusive methods for monitoring HR and RR, in-
cluding a hybrid imaging approach, under both simulated and real driving conditions. The
feasibility of these methods was assessed by comparing data from non-intrusive sensors
with reference sensors. In laboratory settings, magnetic induction and photoplethysmogra-
phy sensors, integrated into seatbelts, along with hybrid imaging combining visual and
thermal data, were tested for RR detection. In real driving conditions, HR and RR were
Sensors 2025,25, 562 8 of 57
monitored using hybrid imaging and a seat-integrated capacitive ECG across both urban
and rural scenarios. The laboratory tests demonstrated reliable RR detection with all three
sensor technologies. In real-world rural driving, both HR and RR were reliably detected.
However, only RR detection is feasible in urban driving due to motion artifacts affecting
the capacitive ECG, which impairs HR detection.
The system proposed in [
37
] allows driver and vehicle monitoring, stating that the
driver affects the vehicle’s performance and vice versa. The proposed system integrates on-
board and remote vehicle sensors to develop algorithms that estimate pollutant emissions,
fuel consumption, driving behavior, and driver health. The main contribution lies in
analyzing the interaction between these factors and between driver behavior and vehicle
performance, conducting an experimental analysis with different drivers and routes, and
the results are implemented in a mobile application. Unlike commercial systems, this
approach utilizes standard sensors and established algorithms tested interactively. During
the system design, principal component analysis was employed to reduce the number
of variables for training and testing, minimizing Bluetooth data transfer between the
biometric wristband, smartphone, and vehicle’s central computer, as shown in Figure 5.
Experimental results indicate the system accurately predicts fuel consumption 84% of the
time, pollutant emissions 89%, and driving behavior 89%. Notably, the study revealed
significant correlations between the driver’s heart condition and traffic conditions.
Sensors 2025, 25, x FOR PEER REVIEW 9 of 62
also functioning through clothing. The model demonstrated very high accuracy: 100% for
high-quality ECG, 96.67% for medium-quality cECG, and 98.08% for lower-quality cECG.
Figure 5. The system proposed in [37] integrates on-board and remote vehicle sensors to develop
algorithms that estimate pollutant emissions, fuel consumption, driving behavior, and driver health.
2.2. On-Board Systems for Driver Drowsiness Monitoring
To mitigate the risk of accidents caused by driver drowsiness, Ebrahimian et al. in-
troduced a novel methodology for multi-level drowsiness detection using Convolutional
Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks [39]. This ap-
proach relies on HR and Respiratory Rate Variations (RRV) data. The authors tested their
model on data from 30 participants in a driving simulation, achieving effective results
with the combined CNN and LSTM models. Specifically, they reported an accuracy of
91% for a three-level drowsiness classification and 67% for a five-level classification.
Similarly, Alguindigue et al. conducted research aimed at reducing drowsy-driving-
related accidents by monitoring sleep-induced aention loss [40]. To assess driver sleepi-
ness, they explored various indicators, including HRV, percentage of eyelid closure over
time (PERCLOS), blink rate and percentage, and EDA signal. They employed three deep
learning (DL) algorithms: a Sequential Neural Network (SNN) for HRV, a 1D-CNN for
EDA, and a Recurrent Convolutional Neural Network (CRNN) for eye-tracking data. The
models using HRV and EDA provided high accuracy, achieving 98.28% and 96.32%, re-
spectively, promoting real-time neuro-adaptive systems to detect driver drowsiness. In
[41], a different approach is used to detect driver drowsiness based on the Johns Drowsi-
ness Scale (JDS). This method utilizes Optalert glasses equipped with an IR transmier/re-
ceiver to measure eyelid velocity and blink duration through reflectance oculography. The
JDS provides ratings on a scale from 1 to 10, derived by the mean and standard deviation
of eyelid movement duration and relative velocity during blinks. According to this, driv-
ers are classified as alert if their score falls between 0 and 4 and drowsy between 5 and 10.
Sukumar et al. proposed a model to determine driver stress and drowsiness by de-
tecting ECG signals, head movements by accelerometers, and hand contact with the steer-
ing wheel by interdigitated capacitive sensors [14]. By integrating these data, the authors
gained insights into driving behavior; to assess stress level, they analyzed ECG signals
using the Synchronized Continuous Wavelet Transform (CWT-SST) to evaluate HRV. The
same algorithm was applied to detect drowsiness level through head movement data.
Their model provided an accuracy of 91.82% in stress detection and 97.30% in identifying
drowsiness level. Amidei et al. developed a model to detect driver drowsiness that pro-
vides timely warnings [42]. Their approach uses the Empativa E4 wristband that monitors
the skin conductance; moreover, the authors tested three ML algorithms for classifying
drowsiness, finding that the RF one performed best, achieving an accuracy of 84.1%.
Figure 5. The system proposed in [
37
] integrates on-board and remote vehicle sensors to develop
algorithms that estimate pollutant emissions, fuel consumption, driving behavior, and driver health.
The model proposed by Škori´c in [
38
] explores the classification of stress levels using
capacitive Electrocardiogram (cECG) signals recorded during driving with non-intrusive
acquisition systems featuring various hardware configurations. The ML model developed
for this purpose relies on four features, all derived from detecting the R peak (the local
maximum of the ECG signal), which is recognized as the most reliably identified point even
in lower-quality cECG recordings. These features were chosen for their low computational
complexity, facilitating real-time application. The model was evaluated using three datasets
of driving-related recordings: high-quality ECG captured with direct skin contact electrodes,
medium-quality cECG obtained via a portable cushion with electrodes operating through
clothing, and lower-quality cECG from car seat-embedded electrodes also functioning
through clothing. The model demonstrated very high accuracy: 100% for high-quality ECG,
96.67% for medium-quality cECG, and 98.08% for lower-quality cECG.
2.2. On-Board Systems for Driver Drowsiness Monitoring
To mitigate the risk of accidents caused by driver drowsiness, Ebrahimian et al. in-
troduced a novel methodology for multi-level drowsiness detection using Convolutional
Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks [
39
]. This ap-
Sensors 2025,25, 562 9 of 57
proach relies on HR and Respiratory Rate Variations (RRV) data. The authors tested their
model on data from 30 participants in a driving simulation, achieving effective results with
the combined CNN and LSTM models. Specifically, they reported an accuracy of 91% for a
three-level drowsiness classification and 67% for a five-level classification.
Similarly, Alguindigue et al. conducted research aimed at reducing drowsy-driving-
related accidents by monitoring sleep-induced attention loss [
40
]. To assess driver sleepi-
ness, they explored various indicators, including HRV, percentage of eyelid closure over
time (PERCLOS), blink rate and percentage, and EDA signal. They employed three deep
learning (DL) algorithms: a Sequential Neural Network (SNN) for HRV, a 1D-CNN for
EDA, and a Recurrent Convolutional Neural Network (CRNN) for eye-tracking data. The
models using HRV and EDA provided high accuracy, achieving 98.28% and 96.32%, respec-
tively, promoting real-time neuro-adaptive systems to detect driver drowsiness. In [
41
],
a different approach is used to detect driver drowsiness based on the Johns Drowsiness
Scale (JDS). This method utilizes Optalert glasses equipped with an IR transmitter/receiver
to measure eyelid velocity and blink duration through reflectance oculography. The JDS
provides ratings on a scale from 1 to 10, derived by the mean and standard deviation of
eyelid movement duration and relative velocity during blinks. According to this, drivers
are classified as alert if their score falls between 0 and 4 and drowsy between 5 and 10.
Sukumar et al. proposed a model to determine driver stress and drowsiness by detecting
ECG signals, head movements by accelerometers, and hand contact with the steering wheel
by interdigitated capacitive sensors [
14
]. By integrating these data, the authors gained insights
into driving behavior; to assess stress level, they analyzed ECG signals using the Synchronized
Continuous Wavelet Transform (CWT-SST) to evaluate HRV. The same algorithm was applied
to detect drowsiness level through head movement data. Their model provided an accuracy of
91.82% in stress detection and 97.30% in identifying drowsiness level. Amidei et al. developed
a model to detect driver drowsiness that provides timely warnings [
42
]. Their approach
uses the Empativa E4 wristband that monitors the skin conductance; moreover, the authors
tested three ML algorithms for classifying drowsiness, finding that the RF one performed best,
achieving an accuracy of 84.1%.
To enhance vehicle and driver safety, Díaz-Santos et al. proposed a model combining
facial recognition and drowsiness detection [
43
], which operates in two phases: the first
identifies the driver before the vehicle starts, and the second monitors eye activity for
fatigue and drowsiness signs (Figure 6). For driver identification, the authors utilized
OpenCV, a computer vision library that detects facial features such as eyes, nose, and
mouth. Drowsiness detection relies on prolonged eye closure by a CNN model. The
system was developed in Python, employing the Keras deep learning framework and
OpenCV. The proposed model achieved 100% recognition accuracy under ideal conditions,
which decreased slightly to 97.5% when the driver wore glasses. Khan et al. implemented
a non-intrusive DMS using computer vision techniques [
44
]. The proposed framework
operates on four levels: an embedded system, edge computing, cloud computing, and
a user interface. In the edge computing layer, facial features such as eye, mouth, and
jaw movements are analyzed to detect drowsiness signs. The system then sends notifi-
cations to relevant authorities, including general information and drowsiness score, to
assess the driver’s condition; after testing, the model provided high robustness with 96%
overall accuracy.
Sensors 2025,25, 562 10 of 57
Sensors 2025, 25, x FOR PEER REVIEW 10 of 62
To enhance vehicle and driver safety, Díaz-Santos et al. proposed a model combining
facial recognition and drowsiness detection [43], which operates in two phases: the first
identifies the driver before the vehicle starts, and the second monitors eye activity for fa-
tigue and drowsiness signs (Figure 6). For driver identification, the authors utilized
OpenCV, a computer vision library that detects facial features such as eyes, nose, and
mouth. Drowsiness detection relies on prolonged eye closure by a CNN model. The sys-
tem was developed in Python, employing the Keras deep learning framework and
OpenCV. The proposed model achieved 100% recognition accuracy under ideal condi-
tions, which decreased slightly to 97.5% when the driver wore glasses. Khan et al. imple-
mented a non-intrusive DMS using computer vision techniques [44]. The proposed frame-
work operates on four levels: an embedded system, edge computing, cloud computing,
and a user interface. In the edge computing layer, facial features such as eye, mouth, and
jaw movements are analyzed to detect drowsiness signs. The system then sends notifica-
tions to relevant authorities, including general information and drowsiness score, to as-
sess the driver’s condition; after testing, the model provided high robustness with 96%
overall accuracy.
Figure 6. Innovative model for facial recognition and driver drowsiness detection proposed in [43].
Safarov et al. in [15] applied a combination of deep learning (DL) and computer vi-
sion algorithms to detect driver drowsiness. They trained their model using custom da-
tasets and tested it with multiple subjects. Eye-blink frequency and mouth region coordi-
nates were captured using facial landmarks (Figure 7). These landmarks were analyzed
in real time to monitor fluctuations in eye-blinking rates and mouth movements. Their
real-time experiments revealed a correlation between yawning and prolonged eye closure,
which were classified as signs of drowsiness. The model demonstrated strong perfor-
mance, achieving 95.8% accuracy for detecting closed eyes, 97% for open eyes, 84% for
yawning, 98% for detecting right-sided head falling, and 100% for left-sided head falling.
Additionally, the system enabled real-time eye state analysis, effectively classifying eyes
into “Open” or “Closed” categories based on threshold values.
The proposed approach in [15] may not be suitable for drivers wearing sunglasses.
Sunglasses can obstruct the model’s ability to accurately detect eye landmarks, which are
crucial for measuring eye blinks and assessing driver alertness. Building on the concept
of non-intrusive monitoring, Kundinger et al. developed an ML approach for detecting
driver drowsiness using physiological data from a sensor embedded in a wrist-worn
bracelet (Figure 8) [45]. They compared the bracelet’s performance against a medical-
grade ECG device. Among the binary classification algorithms tested, the K-Nearest
Neighbor (KNN) model achieved the highest accuracy at 92.13%.
Figure 6. Innovative model for facial recognition and driver drowsiness detection proposed in [43].
Safarov et al. in [
15
] applied a combination of deep learning (DL) and computer
vision algorithms to detect driver drowsiness. They trained their model using custom
datasets and tested it with multiple subjects. Eye-blink frequency and mouth region
coordinates were captured using facial landmarks (Figure 7). These landmarks were
analyzed in real time to monitor fluctuations in eye-blinking rates and mouth movements.
Their real-time experiments revealed a correlation between yawning and prolonged eye
closure, which were classified as signs of drowsiness. The model demonstrated strong
performance, achieving 95.8% accuracy for detecting closed eyes, 97% for open eyes, 84%
for yawning, 98% for detecting right-sided head falling, and 100% for left-sided head falling.
Additionally, the system enabled real-time eye state analysis, effectively classifying eyes
into “Open” or “Closed” categories based on threshold values.
Sensors 2025, 25, x FOR PEER REVIEW 11 of 62
Vyas et al. introduced the “DriverSense” system, designed to enhance the safety of
drivers on two- or three-wheeled vehicles [46]. The framework relies on a wearable device
equipped with mobility sensors (gyroscope, accelerometer, GPS) and biosensors able to
collect real-time biosignals (EEG-electroencephalogram, PPG, ECG, SpO
2
). By combining
the detected signals, the designed system can assess the driver’s stress level and driving
behavior in real time, classifying the first as normal or stressed and the second one into
three categories: slow, normal, and aggressive. The authors reported classification accu-
racies of 86.05% for driving behavior and 88.24% for stress detection.
Figure 7. Main features of the method proposed by the authors in [15]; the Haar–Cascade classifier
was trained to detect faces and extract the related features (a-d). After detection, the authors cap-
tured the coordinates of facial landmarks and exported them to a comma-separated value (csv) file,
based on the proposed classes.
Figure 8. Architecture proposed in [45] for non-intrusive monitoring of the driver to detect the state
of drowsiness by processing the ECG signals via ML algorithms.
2.3. On-Board System for Driver Drunkenness Monitoring
Figure 7. Main features of the method proposed by the authors in [
15
]; the Haar–Cascade classifier
was trained to detect faces and extract the related features (a–d). After detection, the authors captured
the coordinates of facial landmarks and exported them to a comma-separated value (csv) file, based
on the proposed classes.
Sensors 2025,25, 562 11 of 57
The proposed approach in [
15
] may not be suitable for drivers wearing sunglasses.
Sunglasses can obstruct the model’s ability to accurately detect eye landmarks, which are
crucial for measuring eye blinks and assessing driver alertness. Building on the concept of
non-intrusive monitoring, Kundinger et al. developed an ML approach for detecting driver
drowsiness using physiological data from a sensor embedded in a wrist-worn bracelet
(Figure 8) [
45
]. They compared the bracelet’s performance against a medical-grade ECG
device. Among the binary classification algorithms tested, the K-Nearest Neighbor (KNN)
model achieved the highest accuracy at 92.13%.
Sensors 2025, 25, x FOR PEER REVIEW 11 of 62
Vyas et al. introduced the “DriverSense” system, designed to enhance the safety of
drivers on two- or three-wheeled vehicles [46]. The framework relies on a wearable device
equipped with mobility sensors (gyroscope, accelerometer, GPS) and biosensors able to
collect real-time biosignals (EEG-electroencephalogram, PPG, ECG, SpO
2
). By combining
the detected signals, the designed system can assess the driver’s stress level and driving
behavior in real time, classifying the first as normal or stressed and the second one into
three categories: slow, normal, and aggressive. The authors reported classification accu-
racies of 86.05% for driving behavior and 88.24% for stress detection.
Figure 7. Main features of the method proposed by the authors in [15]; the Haar–Cascade classifier
was trained to detect faces and extract the related features (a-d). After detection, the authors cap-
tured the coordinates of facial landmarks and exported them to a comma-separated value (csv) file,
based on the proposed classes.
Figure 8. Architecture proposed in [45] for non-intrusive monitoring of the driver to detect the state
of drowsiness by processing the ECG signals via ML algorithms.
2.3. On-Board System for Driver Drunkenness Monitoring
Figure 8. Architecture proposed in [
45
] for non-intrusive monitoring of the driver to detect the state
of drowsiness by processing the ECG signals via ML algorithms.
Vyas et al. introduced the “DriverSense” system, designed to enhance the safety of
drivers on two- or three-wheeled vehicles [
46
]. The framework relies on a wearable device
equipped with mobility sensors (gyroscope, accelerometer, GPS) and biosensors able to
collect real-time biosignals (EEG-electroencephalogram, PPG, ECG, SpO
2
). By combining
the detected signals, the designed system can assess the driver’s stress level and driving
behavior in real time, classifying the first as normal or stressed and the second one into three
categories: slow, normal, and aggressive. The authors reported classification accuracies of
86.05% for driving behavior and 88.24% for stress detection.
2.3. On-Board System for Driver Drunkenness Monitoring
Drunkenness is a significant cause of traffic accidents. To address this issue, Ku-mar
et al. proposed an IoT-based system designed to detect driver intoxication and prevent the
car engine from starting [
47
]. Their model utilizes advanced sensors to detect alcohol on
the driver’s breath (using an MQ-3 sensor), a Raspberry Pi microcontroller, and a Global
Positioning System (GPS) module for tracking the vehicle’s location. When the driver
attempts to start the car, the system checks for alcohol levels. If the detected level exceeds
legal limits, the microcontroller activates a buzzer, warning the driver that the engine will
shut off shortly. Simultaneously, a Global System Mobile (GSM) module sends the vehicle’s
position to a monitoring center or designated contacts. Similarly, Liu et al. developed an
intoxication detection system integrated into the vehicle’s steering wheel for improved data
accuracy [
48
]. This system uses an array of sensors, as shown in Figure 9, and processes
the data through a weighted fusion algorithm. If the sensors detect signs of intoxication, a
signal is sent to the microcontroller, which then acts on the ignition relay to prevent the
engine from starting.
Sensors 2025,25, 562 12 of 57
Sensors 2025, 25, x FOR PEER REVIEW 12 of 62
Drunkenness is a significant cause of traffic accidents. To address this issue, Ku-mar
et al. proposed an IoT-based system designed to detect driver intoxication and prevent
the car engine from starting [47]. Their model utilizes advanced sensors to detect alcohol
on the driver’s breath (using an MQ-3 sensor), a Raspberry Pi microcontroller, and a
Global Positioning System (GPS) module for tracking the vehicle’s location. When the
driver aempts to start the car, the system checks for alcohol levels. If the detected level
exceeds legal limits, the microcontroller activates a buzzer, warning the driver that the
engine will shut off shortly. Simultaneously, a Global System Mobile (GSM) module sends
the vehicle’s position to a monitoring center or designated contacts. Similarly, Liu et al.
developed an intoxication detection system integrated into the vehicle’s steering wheel
for improved data accuracy [48]. This system uses an array of sensors, as shown in Figure
9, and processes the data through a weighted fusion algorithm. If the sensors detect signs
of intoxication, a signal is sent to the microcontroller, which then acts on the ignition relay
to prevent the engine from starting.
Figure 9. MQ3 gas sensor array integrated into the steering wheel used for the model proposed by
the authors in [48]. They also integrated an Organic Light-Emiing Diode (OLED) display screen
showing the alcohol concentration level, indicating the level of drunkenness.
Swarna et al. proposed an architecture known as the Improved IoT-based Driver
Safety Model (IIoTDSM) to prevent driving under the influence of alcohol [49]. This sys-
tem integrates alcohol detection sensors into the vehicle’s steering wheel. Key components
include a Wi-Fi module, an Arduino microcontroller, an MQ3 gas sensor, and a blink sen-
sor. If the system detects that the driver’s Blood Alcohol Concentration (BAC) exceeds
legal limits, it prevents the vehicle from starting and alerts the appropriate authorities.
In response to the rise of shared car usage, Wang et al. developed a model to detect
intoxicated individuals inside the vehicle or determine if the driver is drunk [50]. Their
system features an electronic nose equipped with gas sensors to measure alcohol concen-
tration in the vehicle’s air. The detection process involves two steps: first, assessing the
overall alcohol level in the car, and second, determining if the driver is intoxicated. These
steps are illustrated in Figure 10. The authors reported high accuracy, achieving 99.44%
in the first and 100% in the second steps, with a sampling time of just five seconds.
Figure 9. MQ3 gas sensor array integrated into the steering wheel used for the model proposed by
the authors in [
48
]. They also integrated an Organic Light-Emitting Diode (OLED) display screen
showing the alcohol concentration level, indicating the level of drunkenness.
Swarna et al. proposed an architecture known as the Improved IoT-based Driver
Safety Model (IIoTDSM) to prevent driving under the influence of alcohol [
49
]. This system
integrates alcohol detection sensors into the vehicle’s steering wheel. Key components
include a Wi-Fi module, an Arduino microcontroller, an MQ3 gas sensor, and a blink sensor.
If the system detects that the driver’s Blood Alcohol Concentration (BAC) exceeds legal
limits, it prevents the vehicle from starting and alerts the appropriate authorities.
In response to the rise of shared car usage, Wang et al. developed a model to detect
intoxicated individuals inside the vehicle or determine if the driver is drunk [
50
]. Their
system features an electronic nose equipped with gas sensors to measure alcohol concen-
tration in the vehicle’s air. The detection process involves two steps: first, assessing the
overall alcohol level in the car, and second, determining if the driver is intoxicated. These
steps are illustrated in Figure 10. The authors reported high accuracy, achieving 99.44% in
the first and 100% in the second steps, with a sampling time of just five seconds.
Sensors 2025, 25, x FOR PEER REVIEW 13 of 62
Figure 10. Flowchart of the model proposed in [50], which checks whether the driver has drunk
alcohol, preventing the engine from being started if the outcome is positive.
In [51], the MQ-3 sensor is used to measure BAC through the driver’s breath and
surrounding air, comparing the readings with the legal limits in respective countries. The
system includes a proximity sensor to prevent the MQ-3 sensor from being covered. If
drunkenness is detected, a flashing red light alerts the driver and the vehicle will not start.
Furthermore, Cho et al. conducted a study using chromatography to detect BAC and as-
sess the drunkenness level [52]. Although this method is more complex, it offers an alter-
native to semiconductor sensors, overcoming the short duration limitations and lack of
stability.
2.4. Comparative Analysis of On-Board Driver Monitoring Systems
The literature analysis has shown that the primary challenge lies in accurately detect-
ing anomalies in driver health, drowsiness, and intoxication using intrusive and non-in-
trusive methods. Various approaches are discussed to assess the driver’s condition, in-
volving monitoring biophysical parameters such as heart rate (HR), Heart Rate Variability
(HRV), Respiratory Rate (RR), and Electrodermal Activity (EDA), and facial expressions
such as blinking frequency, eyelid closure duration, and excessive head movements.
The methods examined in Section 2.3 focus on measuring BAC from breath analysis
by MQ-3 sensors, preventing the engine from starting if the limit is exceeded. Advanced
systems also employ cameras to analyze head movements or facial expressions to enhance
accuracy. Various ML algorithms, such as Random Forest (RF) and Convolutional Neural
Networks (CNN), are commonly used to assess the model accuracy, with CNNs particu-
larly suited for real-time image analysis. The reviewed methodologies demonstrated high
accuracy, indicating their readiness for commercial implementation. A crucial feature of
these systems is maintaining driver comfort, ensuring that monitoring does not interfere
during driving. Non-intrusive approaches using cameras or sensors integrated into belts
or seats are preferred, as they preserve comfort despite requiring higher computational
power. Ongoing research aims to enhance the DMS accuracy in assessing health condi-
tions, for instance by combining biophysical data, postural changes, and driving behavior.
In this regard, inertial sensors on the steering wheel can measure the steering force, piezo-
resistive sensors in the seatbelt can monitor the respiratory activity, and the driver’s pos-
ture can be monitored by pressure sensors in the seat and the stress level by ECG elec-
trodes placed on the steering wheel.
Figure 10. Flowchart of the model proposed in [
50
], which checks whether the driver has drunk
alcohol, preventing the engine from being started if the outcome is positive.
In [
51
], the MQ-3 sensor is used to measure BAC through the driver’s breath and
surrounding air, comparing the readings with the legal limits in respective countries. The
Sensors 2025,25, 562 13 of 57
system includes a proximity sensor to prevent the MQ-3 sensor from being covered. If
drunkenness is detected, a flashing red light alerts the driver and the vehicle will not start.
Furthermore, Cho et al. conducted a study using chromatography to detect BAC and assess
the drunkenness level [
52
]. Although this method is more complex, it offers an alternative
to semiconductor sensors, overcoming the short duration limitations and lack of stability.
2.4. Comparative Analysis of On-Board Driver Monitoring Systems
The literature analysis has shown that the primary challenge lies in accurately detecting
anomalies in driver health, drowsiness, and intoxication using intrusive and non-intrusive
methods. Various approaches are discussed to assess the driver’s condition, involving
monitoring biophysical parameters such as heart rate (HR), Heart Rate Variability (HRV),
Respiratory Rate (RR), and Electrodermal Activity (EDA), and facial expressions such as
blinking frequency, eyelid closure duration, and excessive head movements.
The methods examined in Section 2.3 focus on measuring BAC from breath analysis
by MQ-3 sensors, preventing the engine from starting if the limit is exceeded. Advanced
systems also employ cameras to analyze head movements or facial expressions to enhance
accuracy. Various ML algorithms, such as Random Forest (RF) and Convolutional Neural
Networks (CNN), are commonly used to assess the model accuracy, with CNNs particularly
suited for real-time image analysis. The reviewed methodologies demonstrated high
accuracy, indicating their readiness for commercial implementation. A crucial feature of
these systems is maintaining driver comfort, ensuring that monitoring does not interfere
during driving. Non-intrusive approaches using cameras or sensors integrated into belts or
seats are preferred, as they preserve comfort despite requiring higher computational power.
Ongoing research aims to enhance the DMS accuracy in assessing health conditions, for
instance by combining biophysical data, postural changes, and driving behavior. In this
regard, inertial sensors on the steering wheel can measure the steering force, piezo-resistive
sensors in the seatbelt can monitor the respiratory activity, and the driver’s posture can be
monitored by pressure sensors in the seat and the stress level by ECG electrodes placed on
the steering wheel.
Section 5.1 discusses commercial DMSs that detect drowsiness and distractions, which
are major contributors to accidents. Tables 1–3summarize the key features of the re-
viewed studies, including applications, data acquisition devices, monitored parameters,
intrusiveness levels, alarm systems, ML algorithms used, and accuracy rates.
Sensors 2025,25, 562 14 of 57
Table 1. Main characteristics of driver health monitoring systems in the literature.
Reference Application Acquisition
Device
Detected
Parameters Intrusiveness Alarm
Systems AI Algorithm Accuracy
K. Kaur et al.
[11]Health
monitoring Six unspecified
wearable sensors ECG, BP, BT, HR,
RR, BAC High Yes RF, EC, DT 99.86% (RF)
99.83% (EC)
99.70% (DT)
A. Affanni et al.
[12]Well-being
monitoring Smart necklace HR, blood oxygen
saturation (SpO2)Medium No “Mountaineer
algorithm” N.A.
Z. Gong et al.
[13]Heart rate
monitoring (rPPG) Camera HR (rPPG) Low No LAB algorithm by
SenseTime
Pearson correlation
coefficients: 0.943
(night)
0.906 (day)
Y. Jiao et al.
[34]Fatigue
detection Accelerometers HVR, EDA Low No LGBM, RF 88.70% (LGBM)
85.60% (RF)
L. Leicht et al.
[36]Health
monitoring
MI coil, PPG sensor in
seatbelt, camera,
capacitive ECG sensor
into seat
HR, RR Medium No Pan–Tompkins
algorithm,
FaceLAB 5 N.A.
A.E. Campos-
Ferreira
et al. [37]
Health
monitoring Biometric wristband HR, skin
temperature Medium No Principal
Component
Analysis (PCA) N.A.
T. Škori´c
[38]
Health
monitoring
Direct contact electrodes,
capacitive electrodes in
cushion and car seat ECG, cECG
High (direct
contact), Low
(indirect
contact)
No KNN, SVM, ANN 100 %
96.67 %
98.08 % (1)
(1) The accuracy values refer to the authors’ three ECG signal detection methods, direct contact (100%), and indirect contact (96.67% in the case of sensors embedded into cushion and
98.08% in the case of sensors embedded into car seat); N.A.: Not Available.
Sensors 2025,25, 562 15 of 57
Table 2. Main characteristics of driver drowsiness detection systems in the literature.
Reference Application Acquisition
Device
Detected
Parameters Intrusiveness Alarm
Systems AI Algorithm Accuracy
S. Ebrahimian
et al. [39]Drowsiness
detection eWave32D, thermal
camera HR, RRV Medium No CNN-LSTM 91% (1)
67% (2)
J. Alguindigue
et al. [40]Drowsiness
detection
Smart band, smart glasses
HRV, EDA,
PERCLOS Medium No SNN, 1D-CNN,
CRNN 98.28% (3)
96.32% (4)
A.S.
BaHammam
et al. [41]
Drowsiness
detection (JDS) Optalert glasses Eyelid velocity,
blink duration. Low No OptalertTM
Proprietary pattern
recognition algorithm N.A. (*)
N. Sukumar
et al. [14]
Detection of stress
and drowsiness
levels
Accelerometer, capacitive
sensor ECG, head
movement, IDT Medium No CWT-SST 91.82% (5)
97.30% (6)
A. Amidei et al.
[42]Drowsiness
detection Smart band Skin conductance Medium No RF 84.1%
S. Díaz-Santos
et al. [43]Drowsiness
detection Camera Facial recognition Low No OpenCV
CNN 97.5%
M. A. Khan
et al. [44]Drowsiness
detection Camera Facial features Low No Unspecified 96%
F. Safarov et al.
[15]Drowsiness
detection Camera Blink rate Low Yes Computer vision
algorithm 95.8% (7)
97% (8)
T. Kundinger
et al. [45]Drowsiness
detection Smart band ECG Medium No KNN 92.13%
D. A. Vyas et al.
[46]
Detection of stress
and aggression
level of the driver
Mobility sensors, sensors
for biophysical signals EEG, PPG, ECG,
SpO2High Yes KNN, RF 86.05% (KNN)
88.24% (RF)
(
1
) Three-level drowsiness classification; (
2
) five-level drowsiness classification; (
3
) HRV detection; (
4
) EDA detection; (
5
) stress recognition; (
6
) drowsiness detection; (
7
) drowsy eye
detection; (8) open eye detection; (*) N.A.: Not Available.
Sensors 2025,25, 562 16 of 57
Table 3. Main characteristics of drunkenness detection methods in the literature.
Reference Application Acquisition
Device
Detected
Parameters
Intrusiveness
Alarm
Systems Algorithm Performance
T.S. Kumar
et al. [47]Drunkenness
detection MQ-3 sensor, GPS
module BAC Low Yes Threshold-based
algorithm 85–94%
J. Liu et al. [48]Drunkenness
detection MQ-3 sensor BAC Low Yes Data fusion and
threshold-based
algorithm N.A. (*)
M. Swarna
et al. [49]Drunkenness
detection MQ-3 sensor, GPD
module BAC Low Yes IIoTDSM Efficiency: 98.70%
F. Wang
et al. [50]
Driver and
passenger
drunkenness
detection
Array of 21 gas sensors (
1
)
BAC Low Yes KNN, SVM, RF 99.44% (I Step)
100% (II Step)
J. Akanni
et al. [51]Drunkenness
detection MQ-3 sensor BAC Low Yes Threshold-based
algorithm N.A. (*)
Y. Cho et al. [
52
]
Drunkenness
detection Spectroscopic module BAC Low Yes
Absorbance vs alcohol
characteristic 0.9908 (2)
0.9938 (3)
(1) Reported in Table 1of Ref. [50]; (2) R-squared value for first-order equation; (3) R-squared value for second-order equation;.(*) N.A.: Not Available.
Sensors 2025,25, 562 17 of 57
Future challenges in the field of driver health monitoring concern various technologi-
cal, ethical, and practical aspects to achieve the following goals:
•
High accuracy and reliability of parameter detection by using non-intrusive devices
integrated into the vehicle and wearable devices;
•
Integration with embedded ML/DL models to process large data amounts in real time;
•
The development of systems that work in synergy with vehicle diagnostics, au-
tonomous driving, and traffic management solutions;
•Privacy and high security of user data against cyber-attacks and unauthorized use.
Technological advancements are pushing toward systems implementing multi-modal
analysis for detecting and simultaneously processing driver, vehicle, and traffic data.
Recently several sensing technologies have been proposed in the literature for discretely
detecting the driver’s biophysical parameters, addressing critical aspects of safety and
well-being in the driving context. Flexible piezoelectric sensors offer a promising approach
for real-time posture analysis; their integration into car seats enables continuous monitoring
without being intrusive to detect fatigue or discomfort early on [
53
]. Also, radiofrequency
and optical signal reflection can be exploited to extract driver biophysical parameters;
radar and LiDAR devices have demonstrated capability in measuring respiratory and
cardiac parameters (e.g., HR, HRV), providing accurate and real-time physiological data
while maintaining driver comfort [
54
]. Finally, cameras installed in the vehicle cockpit
can be used to measure heart and respiratory parameters; for example, estimating HR
using images acquired from an iPPG (imaging PPG) camera is a novel and non-invasive
technique, particularly useful in dynamic environments like vehicles, as it avoids direct
contact with the driver [
13
]. These approaches highlight the potential of combining different
sensing modalities to create a robust and comprehensive DMS. Future work could focus on
integrating these technologies into a unified platform, addressing challenges such as data
fusion, reliability in varying conditions, and ensuring user privacy.
3. Smart Monitoring Systems Based on Vehicle Information
In recent years, vehicle safety has become a priority in the automotive industry, leading
vehicle manufacturers to develop and implement advanced technologies for monitoring
and enhancing road safety. These systems protect vehicle occupants while improving inter-
action with the surrounding environment. One key safety feature is the Tire Pressure Moni-
toring System (TPMS) [
55
], which alerts drivers to under- or over-inflated tires, improving
vehicle safety. Additionally, sophisticated systems can assess road conditions in real time,
detecting hazards such as potholes, ice, or other surface irregularities. Driver-assistance
technologies, including stability and braking control, employ sensors and cameras to contin-
uously monitor vehicle performance and the surrounding environment, and automatically
intervene in critical situations, providing a more controlled and safer driving experience.
These advancements are often integrated into broader smart mobility strategies, where
connectivity and data collection are crucial for improving road safety and minimizing
accident risks. Such innovations contribute to a safer, more informed, and more responsible
driving experience. A comprehensive vehicle monitoring model relies on engine and
battery health data, tire pressure and wear, and oil and coolant temperatures, all collected
via the on-board diagnostics (OBD-II) system and then processed by ML algorithms to
assess the vehicle safety level, as illustrated in Figure 11. Furthermore, this information
allows predictive maintenance to guarantee optimal vehicle performance and efficiency.
Sensors 2025,25, 562 18 of 57
Sensors 2025, 25, x FOR PEER REVIEW 18 of 62
3. Smart Monitoring Systems Based on Vehicle Information
In recent years, vehicle safety has become a priority in the automotive industry,
leading vehicle manufacturers to develop and implement advanced technologies for
monitoring and enhancing road safety. These systems protect vehicle occupants while
improving interaction with the surrounding environment. One key safety feature is the
Tire Pressure Monitoring System (TPMS) [55], which alerts drivers to under- or over-
inflated tires, improving vehicle safety. Additionally, sophisticated systems can assess
road conditions in real time, detecting hazards such as potholes, ice, or other surface
irregularities. Driver-assistance technologies, including stability and braking control,
employ sensors and cameras to continuously monitor vehicle performance and the
surrounding environment, and automatically intervene in critical situations, providing a
more controlled and safer driving experience. These advancements are often integrated
into broader smart mobility strategies, where connectivity and data collection are crucial
for improving road safety and minimizing accident risks. Such innovations contribute to
a safer, more informed, and more responsible driving experience. A comprehensive
vehicle monitoring model relies on engine and baery health data, tire pressure and wear,
and oil and coolant temperatures, all collected via the on-board diagnostics (OBD-II)
system and then processed by ML algorithms to assess the vehicle safety level, as
illustrated in Figure 11. Furthermore, this information allows predictive maintenance to
guarantee optimal vehicle performance and efficiency.
Integrating vehicle status information with driver and road conditions can
significantly improve driving safety by creating a holistic and reactive system that can
prevent accidents and optimize risk management. Through vehicle diagnostic tools and
driver monitoring, it is possible to achieve adaptation to a driving style; furthermore, by
detecting fatigue or drowsiness signs, the diagnostic system can reduce the vehicle’s
speed until it stops. Also, automatic adaptation of vehicle parameters can be performed
based on real-time information on road conditions (e.g., slippery pavement, heavy traffic,
or sharp curves) by means of the driving assistance systems, such as traction control and
stability, and by warning the driver to slow down in case of dangerous road conditions or
brake/suspension failures.
Figure 11. Vehicle monitoring based on ML algorithms for processing data acquired by sensors for
detecting and predicting vehicle anomalies.
Figure 11. Vehicle monitoring based on ML algorithms for processing data acquired by sensors for
detecting and predicting vehicle anomalies.
Integrating vehicle status information with driver and road conditions can significantly
improve driving safety by creating a holistic and reactive system that can prevent accidents
and optimize risk management. Through vehicle diagnostic tools and driver monitoring,
it is possible to achieve adaptation to a driving style; furthermore, by detecting fatigue or
drowsiness signs, the diagnostic system can reduce the vehicle’s speed until it stops. Also,
automatic adaptation of vehicle parameters can be performed based on real-time information
on road conditions (e.g., slippery pavement, heavy traffic, or sharp curves) by means of the
driving assistance systems, such as traction control and stability, and by warning the driver to
slow down in case of dangerous road conditions or brake/suspension failures.
3.1. On-Board Smart Monitoring Systems for Vehicle Diagnosis
The first approach for monitoring vehicle conditions uses intelligent diagnostic sys-
tems that detect anomalies and malfunctions before the failure occurs. Yang proposes a
model to enhance fault diagnosis in electric vehicles by a system based on the Controller
Area Network (CAN) bus and diagnostic communication architecture compliant with
the AUTOSAR standard [
18
]. This approach aims to improve software reusability and
the portability of fault diagnosis systems for electric vehicles. By leveraging AUTOSAR,
developers can focus more on functionality rather than system architecture. The system
provides a fault detection response time of 0.0217 s and a detection accuracy of 98.70%.
Kim et al. developed a diagnostic algorithm to effectively identify and localize faults
of a 4WIS4WID vehicle which, employing redundant hardware and fault-tolerant controls,
allows continuous operation even in the presence of faults [
56
]. The limited degrees of
freedom compared to the number of potential failures complicate the data analysis process.
The algorithm uses residual analysis to compare vehicle behavior with sensor data; these
residuals are converted into three-channel images using histogram analysis and logical
functions to train a CNN algorithm to detect and locate faults. The Recursive Least Squares
(RLS) method also classifies fault types and vehicle output in response to detected issues.
Kannan et al. proposed a fault diagnosis system using a PIC microcontroller within
a Predictive Maintenance (PdM) framework, highlighting the crucial role of extending
vehicle life [
57
]. The system collects and processes engine, transmission, and control unit
data, utilizing IoT and ML technologies to monitor vehicle health and accurately identify
Sensors 2025,25, 562 19 of 57
faults. The system performs better than existing approaches by generating detailed fault
information and enabling more effective and proactive vehicle management.
Min et al. developed a framework focused on sensors’ self-diagnosis, as a sensor is
reliable when it works correctly [
58
]. Their approach uses a residual consistency-checking
algorithm to identify and isolate faulty sensors by leveraging their redundancy. The
system employs a Denoising Sparse Autoencoder (DSAE) integrated with a soft threshold-
shrinking block, enhancing feature representation and anomaly detection accuracy. Experi-
mental results confirm that the developed solution effectively detects and isolates sensor
faults, contributing to more robust and reliable monitoring.
To enhance accident prevention, Shermila et al. proposed an automotive black box that
continuously monitors and records data from on-board sensors to uncover the actual causes
of accidents [
59
]. This system provides information, such as speed and acceleration during
and after an accident, to verify if safety devices functioned correctly in case of impact;
it integrates different sensing devices, including collision sensors to detect the accident,
temperature and gas sensors to identify anomalies derived from the impact, and GPS to
determine the vehicle’s position. In case of emergency, sensor data are transmitted to law
enforcement and medical services via GSM and an internet network. Figure 12 illustrates
the system architecture through a block diagram.
Sensors 2025, 25, x FOR PEER REVIEW 20 of 62
Figure 12. Architecture of the vehicle monitoring system proposed in [59].
3.2. Innovative Systems for Detecting Vehicle Instability Behavior While Driving in Adverse
Asphalt Conditions (Icy Asphalt, Uneven Asphalt, and Aquaplaning)
Driving in difficult environmental conditions poses significant risks, especially at
high speeds. Ensuring vehicle stability on icy or wet roads, where the lack of friction
between asphalt and tires leads to the loss of the vehicle’s control, is essential to increase
driver safety. Fichtinger et al. [60] developed a method combining various aquaplaning
detection models, including monitoring of the slip slope variations (similar to
Gustafsson’s work [61]), changes in rolling resistance, wheel spin-up/down detection, and
analysis of wheel speed variance, acceleration, and rolling resistance using CUSUM
(Cumulative Sum Control Chart). Their approach utilizes a minimal set of sensors,
focusing on Electronic Stability Control (ESC) data and driving torque.
Hu et al. propose an advanced system for identifying slippery road conditions using
Connected Vehicle (CV) technology [22]. Their model leverages data from connected
vehicles and employs deep learning algorithms, particularly LSTM networks. These
networks, trained with simulated data from VISSIM software and optimized using
Bayesian algorithms, accurately classify road conditions into dry, snowy, and icy
categories, achieving Success Rates (SRs) of 100%, 99.06%, and 98.02%, respectively. A key
innovation is the system’s ability to improve driving behavior: issuing timely warnings
helps drivers adjust speed and maintain safe distances, reducing potential accidents by
up to 90%. The system effectively detects black ice, a critical factor in preventing
hazardous driving incidents.
Lee et al. [62] developed an advanced black ice detection system for autonomous
vehicles, notoriously difficult to detect, by using CNN algorithm to identify it accurately.
To enhance the system’s performance, the authors applied data augmentation techniques,
training the CNN with images representing various road conditions prone to black ice
formation. The model achieved a detection accuracy of 96%, showing high efficacy to
reduce accidents in icy conditions. Another critical issue affecting urban roads is the
Figure 12. Architecture of the vehicle monitoring system proposed in [59].
3.2. Innovative Systems for Detecting Vehicle Instability Behavior While Driving in Adverse
Asphalt Conditions (Icy Asphalt, Uneven Asphalt, and Aquaplaning)
Driving in difficult environmental conditions poses significant risks, especially at
high speeds. Ensuring vehicle stability on icy or wet roads, where the lack of friction
between asphalt and tires leads to the loss of the vehicle’s control, is essential to increase
driver safety. Fichtinger et al. [
60
] developed a method combining various aquaplaning
detection models, including monitoring of the slip slope variations (similar to Gustafsson’s
work [
61
]), changes in rolling resistance, wheel spin-up/down detection, and analysis of
Sensors 2025,25, 562 20 of 57
wheel speed variance, acceleration, and rolling resistance using CUSUM (Cumulative Sum
Control Chart). Their approach utilizes a minimal set of sensors, focusing on Electronic
Stability Control (ESC) data and driving torque.
Hu et al. propose an advanced system for identifying slippery road conditions using
Connected Vehicle (CV) technology [
22
]. Their model leverages data from connected
vehicles and employs deep learning algorithms, particularly LSTM networks. These
networks, trained with simulated data from VISSIM software and optimized using Bayesian
algorithms, accurately classify road conditions into dry, snowy, and icy categories, achieving
Success Rates (SRs) of 100%, 99.06%, and 98.02%, respectively. A key innovation is the
system’s ability to improve driving behavior: issuing timely warnings helps drivers adjust
speed and maintain safe distances, reducing potential accidents by up to 90%. The system
effectively detects black ice, a critical factor in preventing hazardous driving incidents.
Lee et al. [
62
] developed an advanced black ice detection system for autonomous
vehicles, notoriously difficult to detect, by using CNN algorithm to identify it accurately.
To enhance the system’s performance, the authors applied data augmentation techniques,
training the CNN with images representing various road conditions prone to black ice
formation. The model achieved a detection accuracy of 96%, showing high efficacy to reduce
accidents in icy conditions. Another critical issue affecting urban roads is the presence
of potholes and cracks, which can make driving uncomfortable and damage vehicles.
Bhadraray et al. addressed this problem by creating a model that adjusts vehicle speed
and trajectory to avoid potholes while maintaining the travel route [
63
]. They simulated
the system using a TurtleBot3 Burger robot equipped with an RGB camera and trained the
YOLOv4-Tiny model on a dataset of 2160 images to detect potholes. Additionally, they
used Canny edge detection to identify road boundaries within the region of interest.
Raja et al. introduced the Smart Pothole Avoidance Strategy (SPAS) that uses the Deep
Deterministic Policy Gradient (DDPG) algorithm for dynamic pothole management [
64
].
A key innovation of SPAS is its hybrid recognition model, combining audio and visual
feedback from passengers through a mechanism called Speech and Gesture (HRM-SG). This
interactive feedback refines the DDPG system’s capabilities, enhancing pothole avoidance
accuracy and improving passenger comfort. SPAS optimizes lane changing and speed
adjustments based on sensor data and passenger input, making travel safer and more
comfortable. Experimental results show significant improvements over existing methods
with a 10–15% increase in pothole avoidance accuracy, a 10–12% boost in comfort, and 8–10%
faster convergence. This work highlights advancements in autonomous vehicle technology
and the value of integrating human feedback to optimize navigation performance.
3.3. Smart Systems for Driver Style and Behavior Monitoring
Sensor data acquisition enables the identification of a driver’s behavior, distinguishing
aggressive and irrational styles [
65
]. Nouh et al. introduced SafeDriver, a Dynamic Driver
Profile (DDP) system, which uses historical data on traffic violations and accidents to
classify drivers into four risk levels [
66
]. The goal was to improve driver behavior and road
safety by alerting other vehicles connected through the IoV. The authors derived driver
profiles using data from the Saudi Traffic Points System Regulation (STPSR) and applied
DL techniques to process them. Figure 13 shows the architecture for driver classification.
Sensors 2025,25, 562 21 of 57
Sensors 2025, 25, x FOR PEER REVIEW 21 of 62
presence of potholes and cracks, which can make driving uncomfortable and damage
vehicles. Bhadraray et al. addressed this problem by creating a model that adjusts vehicle
speed and trajectory to avoid potholes while maintaining the travel route [63]. They
simulated the system using a TurtleBot3 Burger robot equipped with an RGB camera and
trained the YOLOv4-Tiny model on a dataset of 2160 images to detect potholes.
Additionally, they used Canny edge detection to identify road boundaries within the
region of interest.
Raja et al. introduced the Smart Pothole Avoidance Strategy (SPAS) that uses the
Deep Deterministic Policy Gradient (DDPG) algorithm for dynamic pothole management
[64]. A key innovation of SPAS is its hybrid recognition model, combining audio and
visual feedback from passengers through a mechanism called Speech and Gesture (HRM-
SG). This interactive feedback refines the DDPG system’s capabilities, enhancing pothole
avoidance accuracy and improving passenger comfort. SPAS optimizes lane changing and
speed adjustments based on sensor data and passenger input, making travel safer and
more comfortable. Experimental results show significant improvements over existing
methods with a 10–15% increase in pothole avoidance accuracy, a 10–12% boost in
comfort, and 8–10% faster convergence. This work highlights advancements in
autonomous vehicle technology and the value of integrating human feedback to optimize
navigation performance.
3.3. Smart Systems for Driver Style and Behavior Monitoring
Sensor data acquisition enables the identification of a driver’s behavior,
distinguishing aggressive and irrational styles [65]. Nouh et al. introduced SafeDriver, a
Dynamic Driver Profile (DDP) system, which uses historical data on traffic violations and
accidents to classify drivers into four risk levels [66]. The goal was to improve driver
behavior and road safety by alerting other vehicles connected through the IoV. The
authors derived driver profiles using data from the Saudi Traffic Points System Regulation
(STPSR) and applied DL techniques to process them. Figure 13 shows the architecture for
driver classification.
Figure 13. Architecture proposed in [66] based on different functional blocks.
Similarly, Mohammed et al. developed a real-time driving style recognition model
using electronic monitoring devices to detect the vehicle speed, engine RPM, coolant
temperature, and geolocation coordinates [9]. These data are transmied to a server,
visualized through a virtual dashboard, and stored as driving history and route records.
Shichkina et al. proposed an algorithm that processes twenty descriptive features to
Figure 13. Architecture proposed in [66] based on different functional blocks.
Similarly, Mohammed et al. developed a real-time driving style recognition model
using electronic monitoring devices to detect the vehicle speed, engine RPM, coolant
temperature, and geolocation coordinates [
9
]. These data are transmitted to a server,
visualized through a virtual dashboard, and stored as driving history and route records.
Shichkina et al. proposed an algorithm that processes twenty descriptive features to
monitor driving behavior [
67
]. By accessing data from the diagnostic system (OBD-II),
the model gathers information on environmental conditions, driver actions, and vehicle
performance, to be analyzed by a Kohonen network, which clusters drivers based on
driving style and associated risk levels. Chhabra et al. developed a federated learning
approach for monitoring driver behavior, where data from connected vehicles enhance
the algorithm performance [
68
]. By using CNN-LSTM and CNN-Bi-LSTM deep learning
models, they analyzed data collected from aboard-smartphone sensors and in-car devices
to classify driver behavior accurately. In another study, Malik et al. focused on analyzing
driving patterns using real-world data gathered via the vehicle OBD port [
69
], employing
hierarchical clustering, k-means, ANN, and multilayer perceptrons to identify driving
behavior patterns and categorize data into meaningful groups. A key innovation is the use
of Inter-Class-ReLU activation function, which strengthened the model’s robustness and
classification accuracy.
Lattanzi et al. explored driver behavior detection using data from on-board sen-
sors [
70
]. By applying Support Vector Machines (SVM) and a feed-forward Neural Network
(NN), they extracted descriptive features to identify driving patterns, enhance road safety,
and prevent accidents caused by risky behaviors. Dong et al. detected fatigue and dis-
tracted driving by vision-based techniques and ML models [
71
]. For fatigue detection, they
used facial feature analysis by the RF algorithm to assess driving conditions; instead, to
classify distracted driving behaviors, the authors implemented a CNN model, obtaining
91% and 97.5% accuracy values, respectively.
3.4. On-Board Innovative System for Tire Monitoring
Vasantharaj et al. proposed an innovative indirect TPMS that leverages existing vehicle
sensors to measure tire pressure indirectly [
55
]. Their model also detects grip conditions
and pressure loss based on speed, enhancing overall performance monitoring.
Màrton et al. introduced an advanced indirect TPMS (iTPMS) that uses modern signal
processing techniques combined with a CNN for detecting pressure-related eigenfrequen-
cies [21]. By integrating wavelet-Fourier transforms with CNN-based pattern recognition,
Sensors 2025,25, 562 22 of 57
their approach demonstrates superior accuracy and efficiency in experiments, marking a
significant step forward in AI-driven TPMS technologies.
Shan et al. presented another indirect TPMS utilizing wheel speed sensors and on-
board hardware [
72
]. They detail a method that processes wheel speed signals through
denoising, rim error filtering, and resampling. By analyzing wheel speed spectra and tire
vibration characteristics, the system accurately identifies pressure-related frequencies.
Huo et al. developed a tire pressure warning system using magnetoelectric wheel
speed sensors to collect and analyze tire speed data [
73
]. This system, based on beam
and frequency techniques, allows early detection of dangerous conditions. Experimental
results show improved accuracy and stability compared to traditional methods. Focusing
on tire–road interaction, M. Pomoni investigated factors influencing tire–asphalt contact
and advocates for the use of intelligent tires [
74
]. These advanced tires enhance driver
perception and responsiveness, improving safety in varying road conditions.
Yu et al. proposed a tire wear monitoring system that goes beyond the traditional
methods [
75
]. Using a magnetic angular velocity sensor operating at 600 Hz and a mini host
device for data transfer, they apply the Pacejka model to analyze tire behavior. Data features
are extracted by a transformer-based encoder followed by classification through a fully
connected layer to define the tires’ wear degree, achieving a 97.77% accuracy, outperforming
the Recurrent Neural Networks (RNNs) and SVMs by 5.13% and 4.75%, respectively, even
at low sampling rates.
3.5. Innovative Emissions Monitoring Systems
Focusing on vehicle emissions control, Anusha et al. introduced a cloud-based model
for monitoring and analyzing pollutant emissions using neural networks (NNs) [
76
]. Their
proposed architecture enables real-time emissions tracking and alerts drivers to emergen-
cies. The system gathers data from air quality monitoring stations, vehicle-installed sensors,
and remote sensing devices. These data undergo preprocessing, including handling miss-
ing values and removing outliers, to create a robust dataset for neural network training.
Once processed, the model extracts key features such as vehicle type, engine specifications,
maximum speed, and road conditions.
Sao et al. present an innovative method to predict emissions of carbon dioxide
(CO
2
), nitrogen oxides (NO
x
), and carbon monoxide (CO) from diesel vehicles using
ANN [
77
]. They evaluate six operational parameters (vehicle speed, engine speed, engine
torque, coolant temperature, air/fuel ratio, and intake airflow) collected via on-board
diagnostics (OBD) as predictors. The study finds that predictive accuracy improves with
additional parameters, though the extent varies depending on which parameters are
included. Moreover, the accuracy differs based on vehicle type, with models featuring
after-treatment devices showing lower predictive precision compared to those without.
Andrade et al. propose a continuous emissions monitoring system using the OBD-II
interface to indirectly measure vehicle emissions [
78
]. To enhance data accuracy, they
developed a soft-sensor approach that processes engine combustion metrics such as fuel
injection and airflow to estimate CO
2
emissions. Two distinct algorithms, each with tailored
mathematical formulations, handle input-specific data. Additionally, an unsupervised
TinyML method removes outliers, improving sensor accuracy without relying on cloud
connectivity. The results confirm the system’s viability, providing emissions measurements
in gCO2/km. Figure 14 illustrates the emissions monitoring model developed by authors.
Sensors 2025,25, 562 23 of 57
Sensors 2025, 25, x FOR PEER REVIEW 23 of 62
Focusing on vehicle emissions control, Anusha et al. introduced a cloud-based model
for monitoring and analyzing pollutant emissions using neural networks (NNs) [76]. Their
proposed architecture enables real-time emissions tracking and alerts drivers to
emergencies. The system gathers data from air quality monitoring stations, vehicle-
installed sensors, and remote sensing devices. These data undergo preprocessing,
including handling missing values and removing outliers, to create a robust dataset for
neural network training. Once processed, the model extracts key features such as vehicle
type, engine specifications, maximum speed, and road conditions.
Sao et al. present an innovative method to predict emissions of carbon dioxide (CO2),
nitrogen oxides (NOx), and carbon monoxide (CO) from diesel vehicles using ANN [77].
They evaluate six operational parameters (vehicle speed, engine speed, engine torque,
coolant temperature, air/fuel ratio, and intake airflow) collected via on-board diagnostics
(OBD) as predictors. The study finds that predictive accuracy improves with additional
parameters, though the extent varies depending on which parameters are included.
Moreover, the accuracy differs based on vehicle type, with models featuring after-
treatment devices showing lower predictive precision compared to those without.
Andrade et al. propose a continuous emissions monitoring system using the OBD-II
interface to indirectly measure vehicle emissions [78]. To enhance data accuracy, they
developed a soft-sensor approach that processes engine combustion metrics such as fuel
injection and airflow to estimate CO2 emissions. Two distinct algorithms, each with
tailored mathematical formulations, handle input-specific data. Additionally, an
unsupervised TinyML method removes outliers, improving sensor accuracy without
relying on cloud connectivity. The results confirm the system’s viability, providing
emissions measurements in gCO2/km. Figure 14 illustrates the emissions monitoring
model developed by authors.
Figure 14. Emissions monitoring architecture proposed in [78].
3.6. Comparative Analysis of Smart Monitoring Systems Based on Vehicular Information
Monitoring vehicle conditions and driver behavior is essential in reducing road
accidents, controlling emissions, and enhancing driver safety. The first step is to establish
a diagnostic architecture connected to vehicle one, by integrating redundant hardware,
ML algorithms capable of solving multiple simultaneous faults, and self-diagnostic
systems for sensors in order to identify faults and enable predictive maintenance. Table 4
highlights the key features of the diagnostic systems analyzed in the literature.
Driving on slippery roads, e.g., covered with ice or water, significantly reduces
driving safety, because the reduced grip can lead to losing vehicle control; thus,
developing systems that can detect hazardous conditions is essential. The systems
discussed in Section 3.2 can identify aquaplaning conditions by detecting wheel slippage
Figure 14. Emissions monitoring architecture proposed in [78].
3.6. Comparative Analysis of Smart Monitoring Systems Based on Vehicular Information
Monitoring vehicle conditions and driver behavior is essential in reducing road acci-
dents, controlling emissions, and enhancing driver safety. The first step is to establish a
diagnostic architecture connected to vehicle one, by integrating redundant hardware, ML
algorithms capable of solving multiple simultaneous faults, and self-diagnostic systems for
sensors in order to identify faults and enable predictive maintenance. Table 4highlights
the key features of the diagnostic systems analyzed in the literature.
Driving on slippery roads, e.g., covered with ice or water, significantly reduces driving
safety, because the reduced grip can lead to losing vehicle control; thus, developing systems
that can detect hazardous conditions is essential. The systems discussed in Section 3.2 can
identify aquaplaning conditions by detecting wheel slippage or icy roads using cameras
and DL algorithms trained on web images and real-time road photos. Additionally, cameras
installed on the vehicle can detect potholes on the road, which, when combined with speed
modulation and trajectory modification systems, contribute to innovative safety solutions.
Table 5summarizes the key features of the systems analyzed.
Monitoring and analyzing driving styles have become essential for enhancing road
safety and optimizing vehicle management. The proposed solutions for monitoring driving
behavior and improving safety are promising, incorporating advanced technologies such
as deep learning (DL), federated learning, and clustering. However, each approach has
limitations that could be addressed through innovations in sensors, data processing, and
optimization of predictive models. Future improvements could focus on leveraging mo-
bile technologies for data collection, integrating more advanced sensors for more precise
detection of driving styles, and evolving ML algorithms to better accommodate various
driver behaviors and environmental factors. Furthermore, the use of a monitoring box,
which records the driving data, communicates the vehicle’s position, and alerts emergency
services in case of accident, as outlined in [
59
], is vital. Table 6summarizes the key features
of the driving style monitoring systems analyzed in the literature.
Recent research on tire pressure monitoring systems (TPMSs) has led to innovative
approaches that enhance vehicle safety, reliability, and operational efficiency by integrating
advanced technologies and intelligent methods. Some proposed models utilize existing
vehicle sensors to detect tire grip and pressure conditions, which, combined with ML
algorithms such as CNNs, offer a valid solution to reduce the need for additional hardware,
also providing accurate results. However, challenges such as computational complexity,
sensitivity to variable driving conditions, and reliability in complex scenarios remain. For
this purpose, it will be necessary to ensure consistent performance in different operating
conditions, reduce hardware requirements for broader accessibility, and enhance integration
in new vehicles. This can be achieved by developing more robust, interference-resistant
Sensors 2025,25, 562 24 of 57
sensors and optimizing AI algorithms in software solutions. Table 7summarizes the main
characteristics of the tire monitoring models discussed in this section.
Recent research on vehicle emissions monitoring and forecasting has shown signif-
icant advancements. The solutions proposed by Anusha et al., Sao et al., and Andrade
et al. represent crucial steps toward more intelligent and responsive emissions monitoring
systems. These approaches stand out for their integration of advanced technologies such as
cloud computing, neural networks, and Machine Learning (ML), which enable real-time
monitoring and accurate forecasting. However, to enhance reliability, scalability, and accu-
racy, certain challenges remain, including input data quality, the complexity of forecasting
models, and compatibility across various vehicle types. Future improvements could focus
on optimizing data models, integrating advanced sensor fusion techniques, and adapting
the system to accommodate a broader range of vehicles and operating conditions, ensuring
more precise and efficient emissions monitoring. Table 8summarizes the key features of
the emissions monitoring models from the reviewed articles.
Section 5.2 examines commercially available devices for monitoring vehicle conditions,
including innovative anti-collision devices and intelligent sensors.
As highlighted in Tables 4–8, most of the systems developed for monitoring vehicle
conditions, emissions, tires, and driving style are connected to the vehicle’s OBD-II that
provides a lot of parameters such as speed, engine temperature, engine speed, transmission
data, and more. When combined with ML methods, these data enhance the effective-
ness of monitoring systems. Additionally, OBD-II enables the detection and diagnosis
of mechanical issues by storing error codes that can identify specific faults, allowing for
timely interventions and reducing the risk of significant damage. A monitoring system that
integrates with OBD-II can also help ensure lower emissions and regulatory compliance. In
terms of driving safety, OBD-II-based systems can provide valuable insights, such as detect-
ing dangerous driving behaviors or responding to emergencies. In summary, connecting to
the OBD-II system offers access to a rich, standardized data source that can greatly enhance
the capabilities and effectiveness of vehicle monitoring systems.
Vehicle monitoring systems face several challenges, concerning technological, opera-
tional, economic, and regulatory aspects, including the following:
•
Providing high accuracy and reliability by sensors used for a long time, comprising
the development of self-diagnosis models to verify their correct functioning;
•
Ensuring the interoperability and integration with other vehicle systems, given the
heterogeneity of technologies and standards between the different manufacturers;
•
The management and processing of large amounts of data to develop predictive
models to identify possible failures, the control of emissions, or tire wear;
•To guarantee data protection from cyber-attacks;
•The containment of system costs.
In the future, it will be necessary to develop systems capable of performing global
monitoring of vehicle conditions, including the prediction of the failure chain, such as
suspension wear due to worn tires. Regarding systems to detect driving style, the future is
complex, involving the development of software solutions to adapt the vehicle parameters
in real time and automatically to driving style by accessing the vehicle diagnostic system
(OBD-II), thus limiting the driver’s control.
Sensors 2025,25, 562 25 of 57
Table 4. Main features of vehicle diagnostic systems reported in Section 3.1.
Reference Application Sensors Vehicular
Parameters
Driving
Parameters
Vehicle OBD-II
Interaction Detection Model Performance
X. G. Yang
et al. [18]Fault diagnosis Unspecified (1) Unspecified (1) Unspecified (1)Yes AUTOSAR Detection rate: 98.70%,
Response time: 0.0217 s
J. Kim et al.
[56]Fault diagnosis 6-D IMU sensor Tire longitudinal and
lateral force Speed,
heading No CNN Accuracy: 99.50 %
Kannan
et al. [57]Fault diagnosis in a
PdM context On-board sensors
(OBDII) Engine data, transmission
data, control unit N.A. Yes LSTM Accuracy: 99.62%
Sensitivity: 100%
Specificity: 100%
Min et al.
[58]Framework on sensor
self-diagnosis Unspecified (2) Unspecified (2) Unspecified (2)No DSAE AUC_ROC: 0.8516
F1-score: 0.6701
J.P. Shermila
et al. [59]Automotive black
box system DHT 111, MQ-2,
SW-18010p Engine
temperature Speed,
acceleration Yes Proprietary
algorithm
Accuracy: 98%
Recall: 90%
Specificity: 85%
Precision: 75%
(
1
) Ref. [
18
] studies a diagnostic software without explicitly referring to specific types of sensors; (
2
) the model proposed in [
58
] is an algorithm for identifying and isolating faulty
sensors, and as such, it does not refer to specific types of sensors.
Table 5. Main features of innovative systems for detecting vehicle instability conditions described in Section 3.2.
Reference Application Sensors Detection Model Performance
A. Fichtinger et al. [60] Aquaplaning detection MARWIS (1)Threshold-based algorithm 1.5 s detection time
J. Hu et al. [22] Slippery road conditions detection Ground speed, wheel speed, ground
acceleration, and wheel acceleration LSTM SR: 100% (dry road), SR: 99.06% (snowy
road), SR: 98.02% (icy road)
H. Lee et al. [62] Black ice detection Camera CNN Accuracy: 96%
S. Bhadraray et al. [63]
Detection of potholes on the road surface
Turtlebot3 Burger robot with RGB
camera YOLOv4-Tiny N.A. (*)
G. Raja et al. [64] Smart pothole avoidance system (SPAS) Camera DDPG and HRM-SG Accuracy: 94.2%
(1) Mobile Advanced Road Weather Information Sensor; (*) N.A.: Not Available.
Sensors 2025,25, 562 26 of 57
Table 6. Main features of systems for driver style and behavior monitoring reported in Section 3.3.
Reference Application Sensors Vehicular
Parameters
Driving
Parameters
Vehicle OBD-II
Interaction Detection Model Performance
R. Nouh et al. [66]Driver risk
classification Speed sensor and
accelerometer - Speed, acceleration No Collaborative Filtering
(CF)
MAE: 88.41%
MAPE: 90.18%
MSE: 95.12%
K. Mohammed et al. [9] Driving style tracking GPS, on-board sensors
(OBD-II)
Engine speed, coolant
temperature,
geolocation coordinates Speed Yes - N.A. (*)
Y. Shichkina et al. [67]Driving style
recognition
Stereo sensor, camera,
on-board sensors
(OBD-II)
Mass airflow, engine
load, intake manifold
pressure
Speed, throttle position,
acceleration Yes Detectron2 N.A. (*)
R. Chhabra et al. [68]Driver behavior
classification Accelerometer,
gyroscope -Acceleration and
angular speed No CNN-LSTM,
CNN-Bi-LSTM CNN-Bi-LSTM: 87%
M. Malik et al. [69]Driving
behavior On-board sensors
(OBD-II)
Engine RPM,
engine load,
coolant temperature
Speed and throttle
position Yes ANN Accuracy: 98.68%
E. Lattanzi et al. [70]Driving
behavior On-board sensors
(OBD-II) Engine speed,
engine load
Speed, brake pedal
pressure, throttle
position, steering
wheel angle
Yes SVM
FFNN Accuracy: 95%
B.T. Dong et al. [71]Fatigue detection,
distracted behavior S3FD - Facial features No RF
CNN
Accuracy: 91% for
fatigue detection, 97.5%
for distracted behavior
(*) N.A.: Not Available.
Sensors 2025,25, 562 27 of 57
Table 7. Main features of the tire monitoring models reported in Section 3.4.
Reference Application Sensors Vehicular
Parameters
Driving
Parameters
Vehicle OBD-II
Interaction Detection Model Performance
A. Vasantharaj et al. [
55
]
TPMS WSS, pressure
sensors Wheel speed - Yes DSAE-ANN Accuracy: 98.69%
Z. Màrton et al. [21] iTPMS WSS Gear shift, sharp turn
detection Acceleration,
deceleration No CNN, HWFT-64 Accuracy: 97.32%
(HWFT-64)
T. Shan et al. [72] iTPMS WSS, on-board
hardware
Vertical and
circumferential
vibrations, wheel speed - No Threshold-based model N.A. (*)
B. Huo et al. [73] TPMS Magnetoelectric WSS Wheel speed - No Threshold-based model
Accuracy:
79.3% (Radius);
95.6% (Frequency);
94.5% (Composite)
H. Yu et al. [75]Tire wear
detection Magnetic angular
velocity sensor, CAN (
1
)
Engine torque, engine
speed, yaw rate,
four-wheel
angular velocity
Acceleration, brake No Transformer model Accuracy: 97.77%
(1) Controller Area Network (CAN); (*) N.A.: Not Available.
Table 8. Main features of the emissions monitoring system reported in Section 3.5.
Reference Application On-Board
Sensors
Off-Board
Sensors
Vehicular
Parameters
Driving
Parameters
Environmental
and Other
Parameters
Vehicle OBD-II
Interaction Detection Model Performance
N. Anusha
et al. [76]
Monitoring and
analysis of
polluting
emissions
Emissions sensors
(particulate, NOx,
CO2), GPS, IoT
devices, OBD-II
Lidar, cameras, air
quality sensors,
weather stations
Vehicle type,
engine size,
emissions level Speed
Temperature,
weather
conditions, traffic
density, road type
Yes NN N.A. (*)
J. Sao et al. [77]Emissions
prediction of
diesel vehicles PEMS (1)Not
considered
Engine torque,
speed, coolant
temperature
fuel/air ratio,
intake air
mass flow
Speed Not considered Yes ANN N.A.
P. Andrade
et al. [78]CO2emissions
estimate
Engine Control
Unit (EUC)
sensors
Not
considered
RPM, intake air
temperature, mass
air flow Speed Not considered Yes TEDA-based
algorithm Volumetric
efficiency: 80%
(1) Portable Emission Measurement System (PEMS); (*) N.A. Not Available.
Sensors 2025,25, 562 28 of 57
4. Environmental, Road, and Traffic Conditions Monitoring
In recent years, Intelligent Transportation Systems (ITSs) have become increasingly
significant in the context of urban mobility and traffic management. Central to this inno-
vative approach is the real-time monitoring of traffic flow and environmental conditions.
By tracking traffic, authorities can gather vital data on vehicle behavior, congestion, and
accidents, enhancing road infrastructure management and ensuring greater safety for road
users [
79
]. Similarly, monitoring environmental factors like air quality and weather con-
ditions is crucial for driver safety and comfort and for promoting sustainable mobility.
Together, these aspects support the development of intelligent solutions that streamline
traffic, reduce emissions, and elevate the urban quality of life. ITSs, therefore, stand as a
key element for future mobility. Figure 15 illustrates how smart cities can enhance traffic
management by integrating vehicles, infrastructure, and pedestrians.
Sensors 2025, 25, x FOR PEER REVIEW 30 of 62
w
Figure 15. Architecture of a smart city highlighting the interconnectivity between vehicles,
infrastructures, and pedestrians within an urban center.
4.1. Smart Systems for Traffic Monitoring
In the context of road monitoring, Lilhore et al. designed and implemented an
Adaptive Traffic Management (ATM) system utilizing IoT technologies and ML
algorithms to tackle issues such as congestion, pollution, and delays in urban
transportation [25]. This system consists of three core components (vehicles,
infrastructure, and events) and applies various scenarios to address diverse transportation
challenges. It employs the DBSCAN clustering method to detect anomalies and
dynamically adjusts traffic lights based on traffic volume. Experimental results indicate
that ATM surpasses traditional traffic management methods by reducing waiting times,
alleviating congestion, and enhancing road safety, thereby contributing to smarter
transportation planning in urban environments.
Saleem et al. developed an intelligent traffic congestion control system, known as
FITCCS-VN, specifically designed for smart city vehicular networks (VNs) [80]. This
system leverages ML techniques to analyze and manage increasing traffic congestion and
accidents. Collecting data on traffic flow and route availability facilitates more efficient
navigation and reduces congestion. The results show that the model achieves 95%
accuracy with a 5% error rate, outperforming previous approaches and offering
innovative services to help drivers optimize traffic flow. Traditional ML algorithms
require the analysis of sensitive user data, posing privacy risks; Federated Learning (FL)
provides a solution by ensuring data security during training. In Ref. [81], the authors
introduced a cooperative edge caching scheme for next-generation vehicular networks
that combines an elastic FL algorithm for personalized content prediction using an
adversarial autoencoder (AAE), and a multiagent deep reinforcement learning (MADRL)-
based approach for collaborative caching among SBSs (Small-Cells Base Stations),
optimizing the cost and efficiency. Experimental results show improved cache hit ratio,
personalized predictions, and reduced operational costs compared to traditional schemes.
Mandal et al. proposed an automated real-time traffic surveillance system employing
Deep Convolutional Neural Networks (DCNNs) and a graphical user interface [82]. Their
system aims to simplify the tasks of human operators in traffic management centers,
enabling quicker and more proactive responses to mitigate accidents and congestion.
Using an extensive video database, the model detects traffic jams, tracks stationary
vehicles, and counts vehicles by applying pixel-level segmentation and object detection
techniques. The results demonstrate that the system performs effectively under various
Figure 15. Architecture of a smart city highlighting the interconnectivity between vehicles, infrastruc-
tures, and pedestrians within an urban center.
Road and traffic conditions have a significant impact on driver and vehicle health,
affecting overall safety and performance; complex roads, heavy traffic, delays, roadworks,
and poor roads can significantly increase levels of stress and fatigue as well as distraction
and nervousness, increasing the risk of accidents. Traffic conditions and poor roads cause
faster vehicle wear (e.g., of tires, suspension, and shock absorbers), as well as continuous
stop-and-go on congested roads causing thermal stress to the engine and increased brake
wear. Difficult road and traffic conditions produce negative effects on the driver (stress,
fatigue) and the vehicle (wear, malfunctions), creating high-risk situations for accidents.
4.1. Smart Systems for Traffic Monitoring
In the context of road monitoring, Lilhore et al. designed and implemented an Adap-
tive Traffic Management (ATM) system utilizing IoT technologies and ML algorithms to
tackle issues such as congestion, pollution, and delays in urban transportation [
25
]. This
system consists of three core components (vehicles, infrastructure, and events) and applies
various scenarios to address diverse transportation challenges. It employs the DBSCAN
clustering method to detect anomalies and dynamically adjusts traffic lights based on traffic
volume. Experimental results indicate that ATM surpasses traditional traffic management
methods by reducing waiting times, alleviating congestion, and enhancing road safety,
thereby contributing to smarter transportation planning in urban environments.
Saleem et al. developed an intelligent traffic congestion control system, known as
FITCCS-VN, specifically designed for smart city vehicular networks (VNs) [
80
]. This system
leverages ML techniques to analyze and manage increasing traffic congestion and accidents.
Collecting data on traffic flow and route availability facilitates more efficient navigation and
reduces congestion. The results show that the model achieves 95% accuracy with a 5% error
rate, outperforming previous approaches and offering innovative services to help drivers
optimize traffic flow. Traditional ML algorithms require the analysis of sensitive user
Sensors 2025,25, 562 29 of 57
data, posing privacy risks; Federated Learning (FL) provides a solution by ensuring data
security during training. In Ref. [
81
], the authors introduced a cooperative edge caching
scheme for next-generation vehicular networks that combines an elastic FL algorithm for
personalized content prediction using an adversarial autoencoder (AAE), and a multiagent
deep reinforcement learning (MADRL)-based approach for collaborative caching among
SBSs (Small-Cells Base Stations), optimizing the cost and efficiency. Experimental results
show improved cache hit ratio, personalized predictions, and reduced operational costs
compared to traditional schemes.
Mandal et al. proposed an automated real-time traffic surveillance system employing
Deep Convolutional Neural Networks (DCNNs) and a graphical user interface [
82
]. Their
system aims to simplify the tasks of human operators in traffic management centers,
enabling quicker and more proactive responses to mitigate accidents and congestion. Using
an extensive video database, the model detects traffic jams, tracks stationary vehicles, and
counts vehicles by applying pixel-level segmentation and object detection techniques. The
results demonstrate that the system performs effectively under various environmental
conditions, maintaining reliability even with reduced visibility or adverse weather.
For traffic management in Riyadh, Humayun et al. propose an intelligent traffic
management system that integrates IoT, cloud computing, 5G, and big data to enhance
real-time traffic efficiency [
83
]. The system gathers data on road conditions and accidents
and then communicates alert messages. This model is structured into three levels. In
the first level, the Information Acquisition one, data are collected via vehicle-mounted
sensors that capture images and roadside infrastructure that monitors vehicle counts. In the
second Network level, information is transmitted over a 5G Wi-Fi network to the cloud for
processing. In the third level, the Application Level, dashboards integrate Google Maps and
smartphone applications to provide real-time traffic updates, aiding drivers in navigation.
Putra et al. introduce an Intelligent Traffic Monitoring System (ITMS) framework
designed to assess drivers’ punctuality in reaching their destination, even when unexpected
events occur [
84
]. Their ITMS leverages IoT technology to measure speed using four types
of sensors: motion, ultrasonic, passive infrared (PIR), and speed sensors. The system
includes four monitoring tools: RFID tags for vehicles, roadside sensors, IP addresses
for vehicle connectivity, and QR codes, which allow easy scanning by readers installed
throughout the city. Sarrab et al. propose an IoT-based system for real-time traffic data
collection and processing [
85
]. This system updates road congestion and accidents through
roadside message units, enabling drivers to adjust their routes accordingly. The proposed
model uses Wi-Fi to connect roadside sensors with a cloud server, ensuring timely and
accurate information dissemination. Also, the authors proposed an algorithm for vehicle
detection and physical length estimation based on magnetic sensors, which detect the
disturbances in the Earth’s magnetic field caused by moving vehicles. A threshold is
applied to identify vehicle presence based on significant magnetic field fluctuations. Two
magnetic sensors are placed at a known distance; the vehicle speed
(vi
) is calculated by
measuring the travel time between two sensors. The vehicle magnetic length (
vmli
) is
derived as follows:
vmli=vi·oti(1)
where
vi
is the vehicle speed and
oti
the occupancy time. The vehicle’s physical length
(
vpli
) is estimated by subtracting the detection zone length (
idz
) from the magnetic length:
vpli=vmli−idz (2)
with:
idz =vi·TA
dep −TA
arr −dA−B(3)
Sensors 2025,25, 562 30 of 57
where
TA
arr
and
TA
dep
are the arrival and departure times, whereas
dA−B
is the distance
between the sensor nodes Aand B.
Barbosa et al. introduce an innovative system called LightSpaN, a CNN-based model
designed for efficient and rapid training [
86
]. The system’s performance is assessed using
SUMO software and real-world IoT data, achieving a remarkable accuracy rate of 99.9%
in identifying emergency vehicles. Their solution improves vehicle detection speed and
significantly reduces waiting and travel times. LightSpaN is trained on images from sources
such as the COCO API, VOC datasets, Google Images, and IMAGINET.
Kheder et al. [
87
] propose enhancements to traffic monitoring systems by integrating
vehicular cloud computing (VCC) and IoT-assisted robotics (IoRT). Their architecture
incorporates IoT sensor nodes and cameras to collect real-time traffic data. Two deep
learning models are implemented: a modified version of LeNet-5 for traffic sign recognition
and Inception-V3 for detecting traffic lights. Additionally, the authors optimize ultrasonic
sensor positioning to prevent accidents. The processed data are transmitted to the cloud
and made accessible via a mobile app for drivers and commuters. Results show the LeNet-5
model achieves 99.12% and 99.78% accuracy, while Inception-V3 reaches 98.6%, surpassing
other traffic monitoring systems significantly.
Dhingra et al. present an integrated fog and cloud computing framework to address
real-time IoT data analytics challenges in urban traffic monitoring and traffic light manage-
ment [
88
]. Since traditional cloud computing can suffer from delays due to data overload
and network congestion, the authors propose using fog computing nodes (miniature com-
puters) to pre-process real-time data from distributed sensors. This method reduces latency
and enhances system responsiveness, leading to more effective management of traffic
congestion and accident detection. Additionally, the system connects to Twitter to send
congestion alerts, blending technological advancements with practical tools to improve
urban mobility. Despite implementing monitoring systems like SAHER in Saudi Arabia,
drivers continue to circumvent penalties and violate traffic laws.
Khan et al. propose an advanced traffic surveillance system leveraging Un-manned
Aerial Vehicles (UAVs) and 5G technology to tackle road accidents [
89
]. The key innovation
lies in the combination of UAVs and 5G, offering a more dynamic and responsive solution
than traditional traffic monitoring systems. Their results indicate that reducing common
traffic violations significantly lowers accident rates, leading to fewer fatalities and injuries.
Figure 16 illustrates the UAV system’s monitoring capabilities.
Sensors 2025, 25, x FOR PEER REVIEW 32 of 62
optimize ultrasonic sensor positioning to prevent accidents. The processed data are
transmied to the cloud and made accessible via a mobile app for drivers and commuters.
Results show the LeNet-5 model achieves 99.12% and 99.78% accuracy, while Inception-
V3 reaches 98.6%, surpassing other traffic monitoring systems significantly.
Dhingra et al. present an integrated fog and cloud computing framework to address
real-time IoT data analytics challenges in urban traffic monitoring and traffic light
management [88]. Since traditional cloud computing can suffer from delays due to data
overload and network congestion, the authors propose using fog computing nodes
(miniature computers) to pre-process real-time data from distributed sensors. This
method reduces latency and enhances system responsiveness, leading to more effective
management of traffic congestion and accident detection. Additionally, the system
connects to Twier to send congestion alerts, blending technological advancements with
practical tools to improve urban mobility. Despite implementing monitoring systems like
SAHER in Saudi Arabia, drivers continue to circumvent penalties and violate traffic laws.
Khan et al. propose an advanced traffic surveillance system leveraging Un-manned
Aerial Vehicles (UAVs) and 5G technology to tackle road accidents [89]. The key
innovation lies in the combination of UAVs and 5G, offering a more dynamic and
responsive solution than traditional traffic monitoring systems. Their results indicate that
reducing common traffic violations significantly lowers accident rates, leading to fewer
fatalities and injuries. Figure 16 illustrates the UAV system’s monitoring capabilities.
Figure 16. UAV system proposed in [89], able to monitor vehicles’ movement by detecting and
highlighting the speeds.
Singh et al. [90] introduce a framework that integrates traffic management with air
quality prediction for smart cities using crowdsourced data. Their approach involves
creating Navigation Reference Spatial Data (NRSD) derived from GPS and
OpenStreetMap trajectories, ensuring spatial reliability. These data feed into a real-time
traffic density analysis utilizing k-means clustering and Graham scanning techniques. For
air quality assessment, the framework employs a Bayesian classifier based on eight
standard parameters. An analysis of three major cities in North India demonstrates 98%
accuracy, reflecting a 3.8% improvement over previous models. Additionally, the system
suggests optimal routes that avoid traffic in 87% of cases. The framework’s innovation lies
in its integrated use of crowdsourced data and advanced spatial–temporal analysis to
enhance real-time traffic management and air quality monitoring.
Figure 16. UAV system proposed in [
89
], able to monitor vehicles’ movement by detecting and
highlighting the speeds.
Sensors 2025,25, 562 31 of 57
Singh et al. [
90
] introduce a framework that integrates traffic management with air
quality prediction for smart cities using crowdsourced data. Their approach involves
creating Navigation Reference Spatial Data (NRSD) derived from GPS and OpenStreetMap
trajectories, ensuring spatial reliability. These data feed into a real-time traffic density
analysis utilizing k-means clustering and Graham scanning techniques. For air quality as-
sessment, the framework employs a Bayesian classifier based on eight standard parameters.
An analysis of three major cities in North India demonstrates 98% accuracy, reflecting a
3.8% improvement over previous models. Additionally, the system suggests optimal routes
that avoid traffic in 87% of cases. The framework’s innovation lies in its integrated use of
crowdsourced data and advanced spatial–temporal analysis to enhance real-time traffic
management and air quality monitoring.
4.2. Off-Board Systems for Road and Environmental Monitoring
For road condition monitoring, Åstrand et al. presented a prototype system for road
condition monitoring in underground mine tunnels in Sweden [
91
]. Their proposed system
aims to improve production efficiency and reduce vehicle wear. The system consists of three
main components. The first is a localizer that uses an extended Rao–Blackwellized particle
filter that combines vehicle sensor data with Wi-Fi signal strength from access points. The
second is road monitoring using two methods, one based on a Kalman filter in synergy with
a vehicle suspension system model and another using road condition measurements based
on power spectral density. Finally, re-routing is planned using a rescheduling algorithm to
make automatic decisions about roads in poor condition, integrating maintenance activities
into the short-term production schedule to minimize operational disruption.
Ye et al. in [
92
] developed an intelligent monitoring system for road pavements
using IoT technology. The project addresses several challenges related to using sensors
for infrastructure monitoring, such as the limited lifespan of sensors, damage caused by
sensor insertion, and managing the large amount of data generated. The proposed system
includes a sensor network, cloud platform, communication, and autonomous power supply
(Figure 17). Specifically, it focuses on cement concrete pavement and integrates sensors with
the material structure to collect data on energy, temperature, humidity, noise, wind, and
vibration. These data support pavement design and maintenance and promote synergistic
applications between vehicles and roads. The authors plan to continue exploring the use of
IoT in road maintenance, safety, and material design.
Sensors 2025, 25, x FOR PEER REVIEW 33 of 62
4.2. Off-Board Systems for Road and Environmental Monitoring
For road condition monitoring, Åstrand et al. presented a prototype system for road
condition monitoring in underground mine tunnels in Sweden [91]. Their proposed
system aims to improve production efficiency and reduce vehicle wear. The system
consists of three main components. The first is a localizer that uses an extended Rao–
Blackwellized particle filter that combines vehicle sensor data with Wi-Fi signal strength
from access points. The second is road monitoring using two methods, one based on a
Kalman filter in synergy with a vehicle suspension system model and another using road
condition measurements based on power spectral density. Finally, re-routing is planned
using a rescheduling algorithm to make automatic decisions about roads in poor
condition, integrating maintenance activities into the short-term production schedule to
minimize operational disruption.
Ye et al. in [92] developed an intelligent monitoring system for road pavements using
IoT technology. The project addresses several challenges related to using sensors for
infrastructure monitoring, such as the limited lifespan of sensors, damage caused by
sensor insertion, and managing the large amount of data generated. The proposed system
includes a sensor network, cloud platform, communication, and autonomous power
supply (Figure 17). Specifically, it focuses on cement concrete pavement and integrates
sensors with the material structure to collect data on energy, temperature, humidity,
noise, wind, and vibration. These data support pavement design and maintenance and
promote synergistic applications between vehicles and roads. The authors plan to
continue exploring the use of IoT in road maintenance, safety, and material design.
Figure 17. Monitoring system for road pavements using IoT technology proposed in [92].
Abdelmalak et al. [93] proposed an innovative method for adaptive and intelligent
pothole detection on asphalt, highlighting the importance of integrated approaches for
efficient repair of road inspection systems. The method, which integrates advanced
software and hardware solutions, has been evaluated by the Pothole Detection Dataset
(PDD) and tested by different computer vision models, including VGG19Net, ResNet-50,
GoogLeNet and AlexNet. Among them, the most effective was AlexNet, with a 92.15%
accuracy, sensitivity of 91.38%, F1 score of 96.52%, and processing time of 279.35 s.
Ye et al. [94] focused on road surface condition monitoring, presenting an IoT-based
architecture to enhance service quality and road infrastructure maintenance. They
developed a vibration monitoring prototype including wireless sensor nodes, a gateway,
a remote server, and a browser interface, and based on multiple layers: data acquisition,
Figure 17. Monitoring system for road pavements using IoT technology proposed in [92].
Abdelmalak et al. [
93
] proposed an innovative method for adaptive and intelligent
pothole detection on asphalt, highlighting the importance of integrated approaches for
efficient repair of road inspection systems. The method, which integrates advanced software
Sensors 2025,25, 562 32 of 57
and hardware solutions, has been evaluated by the Pothole Detection Dataset (PDD) and
tested by different computer vision models, including VGG19Net, ResNet-50, GoogLeNet
and AlexNet. Among them, the most effective was AlexNet, with a 92.15% accuracy,
sensitivity of 91.38%, F1 score of 96.52%, and processing time of 279.35 s.
Ye et al. [
94
] focused on road surface condition monitoring, presenting an IoT-based
architecture to enhance service quality and road infrastructure maintenance. They de-
veloped a vibration monitoring prototype including wireless sensor nodes, a gateway, a
remote server, and a browser interface, and based on multiple layers: data acquisition,
pre-processing, processing, interaction, energy management, and networking. The study
highlights critical aspects such as data pre-processing, wireless communication, and visual-
ization. The prototype enables the use of IoT-based solutions in traffic and environmental
monitoring systems, also improving road maintenance management.
Gaspar et al. [
95
] addressed the growing demand for safe forest roads, due to increased
tourism, climate change, and forest management needs. They proposed enhancing road
infrastructure management through systematic data collection and diagnostic activities by
designing a temperature profile measurement probe and a simulation model to validate the
probe’s effectiveness. This device will be deployed in forest road environments to support
ongoing infrastructure monitoring and maintenance.
Hajder et al. [
96
] developed an automated, maintenance-free system for dosing chem-
ical reagents used in winter road maintenance to combat slipperiness caused by water
crystallization, a problem that significantly reduces friction and increases the accident risk.
The proposed system collects data during salt spreader operations to determine the optimal
reagent density for specific road sections. These data are provided to the thermodynamic
models to calculate the ideal reagent concentration and to NNs to detect anomalies. Unlike
traditional methods, the goal of the proposed system is to enhance application efficiency,
reduce chemical overuse, and minimize the environmental impact.
Chen et al. [
97
] addressed the challenge of road ice, a major traffic safety concern, by
proposing a detection and forecasting system utilizing IoT technology. Their innovation
includes a low-power ice sensor that monitors road conditions and transmits data to an IoT
gateway via LoRa technology. The system features a distributed algorithm on the gateway
that identifies ice formation trends over time, adapting to various road conditions. The al-
gorithm uses road data provided from ice capacitive and temperature sensors, time stamps,
and sensors IDs to monitor the icy road’s risks. It switches between two monitoring modes:
Commfreq_Monitor_Model (low-frequency sampling/transmission) for temperature range
[+2
◦
C
÷
+5
◦
C] and Highfreq_Monitor_Model (high-frequency sampling/transmission) for
temperatures below +2
◦
C. The sensors autonomously switch modes and evaluate icing
risks by analyzing temperature trends and sensor data. Additionally, the authors devel-
oped a deep NN model called Trans-CGAN to provide accurate ice predictions, even with
unbalanced datasets, by combining a time series encoder–decoder structure and generative
adversarial networks (GAN)-based classification. Time-series data are processed with
position encoding and multi-head attention to extract patterns, while GAN components
balance the class distribution by generating positive samples. The proposed multi-head
self-attention is a collection of general self-attention mechanisms:
AttentionOutput =Attention(Q,K,V)(4)
where Q,K, and Vare the matrices constituted by query, key, and value vectors. This ap-
proach improves classification accuracy by leveraging generated and real data to optimize
the model performance, thus enhancing the road safety.
Kotus et al. presented a novel method for detecting wet road surfaces using an acoustic
vector sensor (AVS) [
98
]. Their technique analyzes sound intensity in the frequency domain,
Sensors 2025,25, 562 33 of 57
identifying acoustic events related to vehicle movements. In detail, the following equation
was employed to calculate the sound intensity:
I(ω)=1
2ρrωIm(P1·P∗
2)(5)
where
ρ
is the air density,
r
the spacing between pressure sensors, and
Pi
the Fourier
transform of the pressure
pi
, whereas Im indicates the imaginary parts of a complex
spectrum and asterisks, the complex conjugation.
By calculating the sound direction for different spectral components, the system filters
out irrelevant noise and estimates the road surface condition from the sound intensity
spectrum. In particular, the algorithm identifies acoustic events, such as vehicle sounds,
by detecting increases in sound intensity above background noise, determined by an
exponential filter with long averaging time:
In[k]=α·In[k−1]+(1−α)·In[k−δ](6)
where αis the averaging factor, kthe sample index, and δthe update delay.
Afterward, the algorithm evaluates the frequency spectrum of these sounds, with
higher intensity in specific high-frequency bands (above 2.5 kHz) serving as a clear indicator
of wet road conditions. To ensure reliability, the results are smoothed using filters, reducing
noise and accounting for variations in traffic and environmental conditions. In real tests,
the algorithm demonstrated an accuracy of about 89% in precision, recall, and F1 score. The
study concludes that this technology can be integrated into smart city systems for efficient
road water detection, contributing to safer and more responsive urban environments.
4.3. Comparative Analysis of Traffic Management and Road Condition Monitoring
The traffic management systems proposed in the analyzed literature have demon-
strated excellent performance, utilizing diverse acquisition technologies such as strategi-
cally positioned cameras, ultrasonic sensors, RFID tags, and magnetic sensors. A critical
feature of any effective traffic management model is a seamless interconnection between
vehicles, infrastructure, and users, which helps reduce congestion by providing drivers
with real-time information and suggesting alternative routes to avoid traffic. Such real-time
updates are particularly beneficial for urban and extra-urban road networks, enhancing
overall mobility and driver awareness. An intriguing aspect of the model discussed in [
25
]
is its dynamic traffic light management capability, which improves traffic flow and prevents
queue formation. Effective control of traffic lights within urban centers is essential for
smooth traffic operations and minimizing delays. Another significant advantage of these
systems is their ability to monitor traffic violations (speed limit breaches, dangerous driving,
and other infractions) through strategically placed cameras and sensors, fostering greater
driver responsibility and road safety by reducing accidents. In this regard, Section 5.3
provides an overview of various commercial hardware and software solutions for traffic
control and management. Table 9summarizes the key common features of the systems
proposed in the articles reviewed in Section 4.1.
Monitoring road surface conditions is equally important. Cameras and sensors prop-
erly placed can assess road wear and facilitate predictive maintenance, ensuring safer
driving conditions and extending the road lifespan. Table 10 compares the proposed sys-
tems based on performance metrics, communication technologies, involved entities, and
detection sensors, offering a comprehensive evaluation of their capabilities.
Sensors 2025,25, 562 34 of 57
Table 9. Main features of traffic management systems reported in Section 4.1.
Reference Application IoT Sensors and
Technologies
Communication
Technology
Entities Involved in
Communication
Monitoring
Algorithm Performance
U.K. Lilhore et al. [25]Traffic
management Cameras, RFID tags, GPS
module, WSNs (1)
IP-based internet 3G/4G,
LTE, Wi-Fi, NFC, Bluetooth,
ZigBee
Vehicle, roadside
infrastructure, and events
DBSCAN clustering method
N.A. (*)
M. Saleem
et al. [80]Traffic
management IoV sensors Internet (IoV) Vehicle, roadside
infrastructure FITCCS-VN Accuracy: 95%
Miss rate: 5%
V. Mandal
et al. [82]Traffic
monitoring Cameras N.A. (*) Traffic monitoring centers YOLO
Accuracy: 93.7%
F1-score: 0.8333
RMSE: 154.7741
S3: 0.4034
M. Humayun
et al. [83]Traffic
management Cameras,
magnetic sensors 5G/Wi-Fi Vehicle, roadside
infrastructure, mobile user. Threshold-based method N.A. (*)
A.S. Putra
et al. [84]Traffic
management
Motion, ultrasonic, PIR, and
speed sensor Internet network, cellular
wireless technology Vehicle, mobile user - N.A. (*)
M. Sarrab
et al. [85]Traffic
management Magnetic sensors,
NodeMCU Wi-Fi Roadside
infrastructure Threshold-based method Functionality
evaluation: 80%
R. Barbosa
et al. [86]Traffic monitoring and
vehicle identification Image sensor 5G Roadside
infrastructure Light-SpaN Performance: 99.9%
M.Q. Kheder et al. [87]Traffic monitoring systems
with VCC IR, GPS, ultrasonic sensor,
cameras Wi-Fi, Bluetooth,
cellular networks Vehicle, commuters,
Google’s firebase LeNet-5
Inception-V3 Performance: 99.7%
Performance: 98.6%
S. Dhingra
et al. [88]
Urban traffic
monitoring and light
management HC-SR04 Ultrasonic Sensor Wi-Fi, BLE Roadside
infrastructure Cloud
computing N.A. (*)
N.A. Khan
et al. [89]Traffic surveillance system
based on UAVs Cameras, GPS module 5G UAV, driver, base station Threshold-based method N.A. (*)
S. Singh et al. [90]Navigation reference
spatial data Open street map,
GPS module Unspecified Roadside
infrastructure Bayesian
classifier Accuracy: 98%
Performance: 87%
(1) Wireless Sensor Nodes (WSNs); (*) N.A. Not Available.
Sensors 2025,25, 562 35 of 57
Table 10. Main features of environmental and road condition monitoring systems reported in Section 4.2.
Reference Application IoT Sensors and
Technologies
Communication
Technology
Entities Involved in
Communication Monitoring Algorithm Performance
M. Åstrand et al. [91]Road condition
monitoring Inertial and wheel
speed sensors Wi-Fi WiFi access points Kalman filter-based model N.A. (*)
Z. Ye et al.
[92]
Vehicle type, speed, and
weight; pavement health
status and
environment monitoring
Vibration, temperature,
humidity, and wind sensors,
cameras 5G Roadside infrastructure (5G
gateway), remote server Threshold-based model ±
1 cm localization accuracy,
100 % survival rate
M.E.S. Abdelmalak
et al. [93]Pothole detection on
asphalted roads
Cameras, IMU, GPS module
LoRa
Unmanned ground vehicles
(AGV) and ground station AlexNet Accuracy: 92.15%
Sensitivity: 91.38%
F1-score: 96.52%
Z. Ye et al. [94] Road monitoring system WSNs (acceleration sensing
nodes and gateway) LoraWand, UDP, HTTP
communication protocol Front-end devices and
back-end platform Time features processing Vehicle speed and
whee-lbase error < 2%
G. Gaspar et al. [95]Road infrastructure
management Temperature sensors Generic wireless
transmission technology DBAR logger,
server, database N.A. N.A. (*)
M. Hajder et al. [96] Road maintenance Temperature sensors,
humidity sensors 3G and 4G/LTE
communication Unspecified LSTM Accuracy: 89%
Z. Chen et al.
[97]Road icing detection
and prediction
Ice detection
sensor, DS18B20
temperature sensor
Wireless communication
based on Semtech SX1278 Roadside infrastructure
Position encoding
multi-head attention
mechanism, GAN-based
classification
Precision: 0.743
Recall: 0.811
F1-score: 0.767
J. Kotus et al.
[98]Wet road detection Acoustic vector sensor Unspecified Roadside infrastructure Threshold-based model,
spectral analysis Accuracy: 89%
(*) N.A. Not Available.
Sensors 2025,25, 562 36 of 57
Tables 9and 10 highlight that not all the articles analyzed consider the critical role of
communication between vehicles and infrastructure (V2I), which is essential in the context
of smart cities. First, V2I improves traffic management by enabling better coordination of
traffic lights and faster adaptation to real-time road conditions; it also allows the reduction
of waiting times, alleviation of congestion, and improvement of vehicle energy efficiency,
leading to lower CO
2
emissions. Second, V2I enhances road safety by enabling vital infor-
mation exchange, such as weather alerts or accident reports, allowing drivers to adjust their
behavior and avoid potential accidents. Additionally, V2I is key to integrating autonomous
vehicles with smart infrastructures into urban environments, enabling route optimization
and better interoperability, making cities more accessible, and reducing parking needs. In
conclusion, V2I is a technological advancement essential in transforming cities into smarter,
more livable places that can address mobility and environmental challenges.
Current technologies for traffic monitoring and management, although advanced,
have some limitations that hinder their effectiveness, concerning technological aspects
related to data accuracy, limited coverage on the road network, and fragmented data caused
by discontinuous communication between different monitoring platforms. The growing
amount of data coming from sensors, connected vehicles, and mobile devices requires high
computing power and sophisticated algorithms to support real-time analysis; furthermore,
the poor quality of data (noisy or incomplete data) can lead to errors and ineffective decision
management. Moreover, connected monitoring systems can be vulnerable to cyber-attacks,
making data privacy an ongoing challenge. The enhancement of infrastructures with
advanced sensors (better performance and resistance to environmental conditions), the big
data techniques, predictive analytics, and the adoption of digital twin models are emerging
technologies that offer new opportunities for traffic monitoring and management. A further
challenge is that non-centralized and distributed systems may not be able to determine
alternative routes based on information about the level of traffic.
5. Cutting-Edge Commercial Solutions for Monitoring Vehicle, Driver,
and Traffic Conditions
This section describes the most recent commercial hardware and software solutions
in the field of advanced driver and vehicle monitoring systems, as well as out-of-vehicle
solutions for monitoring traffic, environmental, and road conditions.
5.1. Overview of Commercial Solutions for Driver Condition Monitoring
Many companies are focusing on designing a DMS based on cameras or sensors to
monitor driver behavior and alertness. By processing in-cockpit data in real time through
advanced algorithms, the DMS detects signs of distraction or drowsiness (main causes of
accidents), preserving driver privacy by processing data locally. In May 2024, Smart Eye
Co. (Göteborg, Sweden) introduced the Smart Eye Pro 12 eye-tracking system, featuring
Profile ID and advanced drowsiness detection capabilities to enhance safety and the user
experience (Figure 18) [
99
]. The Profile ID function assigns each user a unique identifier
based on facial features, removing the need for manual identification. The drowsiness
detection technology uses algorithms to assess it on a scale from 4 to 9, in line with
upcoming 2026 regulations mandating such features for vehicle safety. The Smart Eye
Pro 12 release also includes gaze accuracy and pupil tracking for heightened robustness
and reliability. The system features a multi-camera setup, with two to eight cameras to be
placed horizontally or vertically, so enabling 360-degree head and eye tracking. AI-based
emotion software developed by Affectiva Co. (Boston, MA, USA) allows the Smart Eye’s
DMS to capture a variety of features in real time [99]:
Sensors 2025,25, 562 37 of 57
1.
Driver identification: it recognizes the driver to adjust the vehicle’s settings accord-
ingly and ensures that only authorized drivers can use the car.
2.
Distraction, drowsiness, and attention: it monitors the driver’s eye, head, and facial
movements to detect distraction and catch early signs of fatigue, ensuring attention
remains on the road.
3.
Dangerous behavior: it identifies distracting behaviors such as eating, drinking,
smoking, or using a mobile phone.
4.
Object detection: it detects objects within the vehicle and monitors how the driver
interacts with them.
5.
Activity detection: it monitors the driver’s activities to determine if they are focused
on driving or engaged in other tasks.
6.
Body posture: it monitors the driver’s posture, movements, and interactions with
objects or vehicle interfaces.
7.
Facial expression analysis: Affectiva’s Emotion AI interprets the driver’s facial expres-
sions to assess his/her mood, emotions, and behaviors.
8.
Health status: it assesses the driver’s health status by analyzing body posture along
with eye, head, and facial movements.
Sensors 2025, 25, x FOR PEER REVIEW 40 of 62
Figure 18. Key features of the “Smart Eye Pro 12” DMS released in 2024 by Smart Eye company [99].
(a) (b)
(c) (d)
Figure 19. (a) Interior camera system enabling driver and vehicle to interact seamlessly; (b)
developed technologies and detection capabilities from Continental Engineering; (c) hardware and
Figure 18. Key features of the “Smart Eye Pro 12” DMS released in 2024 by Smart Eye company [
99
].
Recently the Continental Engineering Co. (Frankfurt, Germany) proposed the “Cabin
Sensing Solution” with technologies and detection capabilities, including a DMS, radar,
cockpit monitoring system, and thermal camera (Figure 19a,b,d) [
100
]. The DMS detects the
driver’s drowsiness and distraction, and supports facial recognition, emotion detection, and
advanced controls (Figure 19c). Its main component is the DMS software algorithm, which
operates on dedicated hardware by a CAN-FD interface or is integrated into a third-party
High-Performance Computer (HPC). A 1MP camera with LED illumination is available as
an encapsulated unit or integrated within the instrument cluster [100].
Sensors 2025,25, 562 38 of 57
Sensors 2025, 25, x FOR PEER REVIEW 40 of 62
Figure 18. Key features of the “Smart Eye Pro 12” DMS released in 2024 by Smart Eye company [99].
(a) (b)
(c) (d)
Figure 19. (a) Interior camera system enabling driver and vehicle to interact seamlessly; (b)
developed technologies and detection capabilities from Continental Engineering; (c) hardware and
Figure 19. (a) Interior camera system enabling driver and vehicle to interact seamlessly; (b) developed
technologies and detection capabilities from Continental Engineering; (c) hardware and software
solutions for face gesture detection; (d) key features of “Cabin Sensing” solution and standards
met [100].
The interior radar enables the child presence detection (CPD) in compliance with the
EuroNCAP 2024 standard (more details at the website https://cdn.euroncap.com/en/
(accessed on 16 December 2024)). In addition, it tracks body movements and vital signs,
including RR and HR, to detect seat occupancy and any children [
100
]. The monitoring
system covers the driver, all passengers, and the full cabin, detecting the presence of
passengers and objects, seat occupancy and seatbelt use, and assessing the occupants’
posture and status for Level 4 and 5 vehicles. For example, it checks if the driver is focused
and their hands are on the wheel. Also, the system includes driver identification (Driver ID)
for personalization and support for video conferencing [
100
]. The thermal camera measures
the temperature of the head, forehead, and body, calculating a Thermal Comfort Value
(TCV) for both driver and passengers. This numeric scale reflects their thermal sensation,
from very cold to very hot, and can be valuable input for optimizing climate control [
100
].
Magna International Inc. (Aurora, ON, Canada) recently launched a Driver and
Occupant Monitoring System (DOMS) that uses ADAS cameras and interior mirrors to
detect driver distraction, drowsiness, and fatigue, aiming to reduce driving distractions
and thus crash risks (Figure 20). The Magna DOMS, integrated into the rearview mirror,
analyses the driver’s head, eye, and body movements to detect signs of distraction or
fatigue. By using cameras and infrared sensors, it can detect the loss of concentration
of the driver, warning him by visual or audible alerts [
101
]. In particular, the proposed
Sensors 2025,25, 562 39 of 57
technology is able to distinguish between routine driving actions, such as checking the
side mirrors, and real distraction actions; once the system recognizes them, it alerts the
driver with acoustic or visual signals. A camera in the interior rearview mirror allows
monitoring of the driver but also other occupants. This system can include features like
child presence detection, seatbelt use verification, and passenger identification, enabling
personalized settings based on user preferences [
101
]. Infrared ethanol sensors positioned
near the driver monitor the alcohol and CO
2
levels in the driver’s diluted exhalations to
determine any BAC level above the legal limit. Magna meets the Euro NCAP and General
Safety Regulation requirements [101].
Sensors 2025, 25, x FOR PEER REVIEW 41 of 62
software solutions for face gesture detection; (d) key features of “Cabin Sensing” solution and
standards met [100].
Magna International Inc. (Aurora, ON, Canada) recently launched a Driver and
Occupant Monitoring System (DOMS) that uses ADAS cameras and interior mirrors to
detect driver distraction, drowsiness, and fatigue, aiming to reduce driving distractions
and thus crash risks (Figure 20). The Magna DOMS, integrated into the rearview mirror,
analyses the driver’s head, eye, and body movements to detect signs of distraction or
fatigue. By using cameras and infrared sensors, it can detect the loss of concentration of
the driver, warning him by visual or audible alerts [101]. In particular, the proposed
technology is able to distinguish between routine driving actions, such as checking the
side mirrors, and real distraction actions; once the system recognizes them, it alerts the
driver with acoustic or visual signals. A camera in the interior rearview mirror allows
monitoring of the driver but also other occupants. This system can include features like
child presence detection, seatbelt use verification, and passenger identification, enabling
personalized seings based on user preferences [101]. Infrared ethanol sensors positioned
near the driver monitor the alcohol and CO
2
levels in the driver’s diluted exhalations to
determine any BAC level above the legal limit. Magna meets the Euro NCAP and General
Safety Regulation requirements [101].
Figure 20. Key features of Magna DMS: distracted driver detection, drowsy detection, occupant
detection, child presence/seat detection, occupant classification, properly worn seatbelt detection.
The last system, proposed by Speedir Inc. (Santee, CA, USA), is an AI dash cam with
eye-tracking functionality to detect drowsy driving. It currently operates as a standalone
tool and does not integrate with typical fleet management platforms. The Driver Fatigue
Monitoring System (DFMS) features a solution for preventing drowsy driving [102];
equipped with a night vision camera and AI facial recognition algorithms, this system
tracks signs of drowsiness and distraction using infrared sensors to monitor the eye
positioning (blink rate, closed eyes, eye direction), head movements (e.g., nodding off),
and yawning, even when the driver is in low light or wearing glasses. Speedir’s DMS is
powered by pre-trained AI software, requiring no Wi-Fi for operation or updates (Figure
21). This system enhances driver safety by alerting users to fatigue or distraction. The key
features of the system are as follows:
Figure 20. Key features of Magna DMS: distracted driver detection, drowsy detection, occupant
detection, child presence/seat detection, occupant classification, properly worn seatbelt detection.
The last system, proposed by Speedir Inc. (Santee, CA, USA), is an AI dash cam
with eye-tracking functionality to detect drowsy driving. It currently operates as a stan-
dalone tool and does not integrate with typical fleet management platforms. The Driver
Fatigue Monitoring System (DFMS) features a solution for preventing drowsy driving [
102
];
equipped with a night vision camera and AI facial recognition algorithms, this system tracks
signs of drowsiness and distraction using infrared sensors to monitor the eye positioning
(blink rate, closed eyes, eye direction), head movements (e.g., nodding off), and yawning,
even when the driver is in low light or wearing glasses. Speedir’s DMS is powered by
pre-trained AI software, requiring no Wi-Fi for operation or updates (Figure 21). This
system enhances driver safety by alerting users to fatigue or distraction. The key features
of the system are as follows:
•
Utilizing AI algorithms, Speedir’s DMS detects drowsy or distracted driving by track-
ing eye and head movements, issuing audible alerts for cell phone use, falling asleep,
or other distractions.
•
With infrared technology, the system performs day or night, unaffected by reflective
light or any glasses the driver may wear.
•
The DMS is designed for universal plug-and-play USB installation, fitting seamlessly
in any vehicle.
•Privacy is protected as it operates without internet or cloud connectivity.
Sensors 2025,25, 562 40 of 57
Sensors 2025, 25, x FOR PEER REVIEW 42 of 62
• Utilizing AI algorithms, Speedir’s DMS detects drowsy or distracted driving by
tracking eye and head movements, issuing audible alerts for cell phone use, falling
asleep, or other distractions.
• With infrared technology, the system performs day or night, unaffected by reflective
light or any glasses the driver may wear.
• The DMS is designed for universal plug-and-play USB installation, fiing seamlessly
in any vehicle.
• Privacy is protected as it operates without internet or cloud connectivity.
Figure 21. Driver Fatigue Monitoring System for driver’s safety by Speedir [102].
Table 11 compares the commercial DMSs previously analyzed, considering their
detection capabilities, integrated sensing devices, monitored human body and car parts,
and developed detection algorithms, offering a comprehensive evaluation of
potentialities.
Table 11. Comparison of commercial DMSs illustrated in Section 5.1.
Commercial DMS Detection
Capabilities
Sensing
Devices
Body and Vehicle Parts
Monitored
Installation
Area
Detection
Algorithm
Smart Eye Pro 12 by
SmartEye [99]
Driver identification,
distraction, drowsiness,
dangerous behaviors,
body posture, facial
expressions
1 MP cameras
Body posture, head and
eye position, eye gaze
and blinking, pupil
diameter
Front of the
cockpit
AI-based
emotion
Cabin Sensing
Solution by
Continental
Engineering [100]
Facial expressions and
recognition, driver
distraction, drowsiness,
head and forehead
temperature, RR, HR
1 MP camera,
radar, thermal
camera
Head and eye position,
seatbelt, seat occupancy
Front of the
cockpit
Proprietary
software
Driver and Occupant
Monitoring System
by Magna
International Inc.
[101]
Driver identification,
distraction, drowsiness,
fatigue, facial expressions,
ethanol environmental
concentration
1.7 MP camera,
IR movement
and ethanol
sensors
Head and eye position,
seatbelt, seat occupancy
Rear-view
mirror and
steering block
Proprietary
software
Driver Fatigue
Monitoring System
by Speedir Inc. [102]
Driver distraction,
drowsiness, facial
expressions
Night vision
camera, IR
sensors
Head and eye position,
eye gaze and blinking,
pupil diameter
Front of the
cockpit
Pre-trained
AI software
5.2. Overview of Commercial Solutions for Monitoring Vehicle Conditions
Figure 21. Driver Fatigue Monitoring System for driver’s safety by Speedir [102].
Table 11 compares the commercial DMSs previously analyzed, considering their
detection capabilities, integrated sensing devices, monitored human body and car parts,
and developed detection algorithms, offering a comprehensive evaluation of potentialities.
Table 11. Comparison of commercial DMSs illustrated in Section 5.1.
Commercial DMS Detection
Capabilities Sensing Devices Body and Vehicle
Parts Monitored Installation Area Detection
Algorithm
Smart Eye Pro 12 by
SmartEye [99]
Driver identification,
distraction,
drowsiness,
dangerous behaviors,
body posture, facial
expressions
1 MP cameras
Body posture, head
and eye position, eye
gaze and blinking,
pupil diameter
Front of the cockpit AI-based emotion
Cabin Sensing
Solution by
Continental
Engineering [100]
Facial expressions
and recognition,
driver distraction,
drowsiness, head and
forehead
temperature, RR, HR
1 MP camera,
radar, thermal
camera
Head and eye
position, seatbelt,
seat occupancy Front of the cockpit Proprietary software
Driver and Occupant
Monitoring System
by Magna
International Inc.
[101]
Driver identification,
distraction,
drowsiness, fatigue,
facial expressions,
ethanol
environmental
concentration
1.7 MP camera, IR
movement and
ethanol sensors
Head and eye
position, seatbelt,
seat occupancy
Rear-view mirror and
steering block Proprietary software
Driver Fatigue
Monitoring System
by Speedir Inc. [102]
Driver distraction,
drowsiness, facial
expressions
Night vision camera,
IR
sensors
Head and eye
position, eye gaze
and blinking, pupil
diameter
Front of the cockpit Pre-trained AI
software
5.2. Overview of Commercial Solutions for Monitoring Vehicle Conditions
An In-Vehicle Monitoring System (IVMS) consists of one or more electronic devices
and purpose-designed software installed in a vehicle to capture and record information
about vehicle performance and driver behavior. Often referred to as the “black box” of the
automotive world, an IVMS typically logs live data such as speed, acceleration, braking
patterns, seatbelt usage, and more. These data provide a picture of activities across a
single vehicle or fleet of vehicles. Moreover, a robust IVMS evaluates driver and vehicle
performance against pre-defined safety standards and criteria. IVMSs range in complexity:
some connect via the vehicle’s OBD-II port for easy plug-and-play functionality, while
others require installation and enable real-time tracking and data downloads. IVMSs
offer a wide range of advantages that enhance the safety, efficiency of vehicles, and cost
management in the case of the fleet of vehicles:
Sensors 2025,25, 562 41 of 57
•
Enhanced driver safety: improve safety by providing feedback on unsafe driving behaviors.
•
Improved driving behavior: discourage unsafe actions like speeding, harsh accelera-
tion, or abrupt braking.
•
Crash detection and response: this allows identification and a quick reaction to acci-
dents, ensuring timely assistance to distressed drivers.
•
Vehicle performance tracking: monitor fuel consumption, mileage, engine health, and
other crucial factors to identify potential issues before they become costly problems,
extending the lifespan of vehicles.
•
Compliance with health and safety standards: help employers meet legal obligations
by monitoring driver hours and addressing fatigue, demonstrating a commitment to
safety and duty of care.
•
Fewer incidents and crashes: reduce accidents by raising awareness of risk factors and
improving driving habits.
•
Lower insurance premium: offer detailed insights into incidents, enabling efficient
claims handling and accurate allocation of responsibility (many insurers provide
premium discounts for fleets using IVMS).
•
Enhanced load and vehicle security: help protect cargo and vehicles from theft or
tampering by tracking their locations in real time.
•
Reduced fuel costs and environmental impact: lower fuel expenses and maintenance
costs by providing insights into fuel consumption, promoting efficient routing, and
reducing idling.
•
Live GPS tracking and geo-fencing: GPS tracking provides an overview of vehicle
locations, and, at the same time, geo-fencing enables the setting of boundaries for vehicle
usage, such as speed limits in specific areas or restricting access to certain zones.
Many companies are focusing on developing new IVMSs based on radar, cameras, GPS,
or multi-sensor setups. In the following section, some cutting-edge commercial solutions
are reported. Trakm8 is a technology company (Birmingham, UK) focused on automotive
telematics solutions and connected vehicles, including fleet cameras, multi-camera systems,
dash cams, and related software. The 4G RH600 camera offers telematics capabilities,
including driver behavior scoring, vehicle health alerts, driver ID, and upcoming ADAS
features to enhance road safety and reduce risks (Figure 22a). Connectivity technical
specifications are Bluetooth Low Energy (BLE 4.2), GPS, 4G technology, vehicle CAN 1,
and Tacho CAN 2 interfaces [
103
]. The RH600 camera (manufactured by Trakm8 Holdings
PLC, Birmingham, UK) can be used as a traditional dash cam in two different ways, with
a single lens facing the road/driver or a dual lens facing both the road and driver. The
RH600 features the Trakm8’s ConnectedCare vehicle health software solution, designed
to reduce breakdowns and non-start risks, manage service and repair costs, and enhance
residual vehicle value. Trakm8 Insight is a software platform that processes data from the
RH600 camera and displays it via a desktop portal or mobile app. The RH600 camera will
implement ADAS features, including driver distraction and fatigue detection, forward
collision warnings, and following distance warnings.
Trakm8’s technology also integrates the RoadHawk DC-4 forward-facing dash cam
(Figure 22b), equipped with Wi-Fi, GPS, and a 1440p Quad HD (2K) camera (30 fps frame
rate), which can capture road signs and license plate numbers, so allowing drivers to save
photos and videos via a gesture sensor. Moreover, it also features a built-in G-Sensor that
automatically allows drivers to save road accident video for later review [
104
]. The systems
designed by Trakm8 also include the RH800 4G Mobile Digital Video Recorder (MDVR),
which supports up to four cameras and features telematics functionality, including GPS
tracking, driver behavior, and vehicle health monitoring (Figure 22c) [
105
]. The Trakm8
Insight telematics system can measure speed, over-revving, and harsh acceleration and
Sensors 2025,25, 562 42 of 57
braking. Trakm8 Insight telematic utilizes GPS, fuel consumption, driving trends, and dash
cam data to offer information supporting management processes.
Sensors 2025, 25, x FOR PEER REVIEW 44 of 62
service and repair costs, and enhance residual vehicle value. Trakm8 Insight is a software
platform that processes data from the RH600 camera and displays it via a desktop portal
or mobile app. The RH600 camera will implement ADAS features, including driver
distraction and fatigue detection, forward collision warnings, and following distance
warnings.
Trakm8’s technology also integrates the RoadHawk DC-4 forward-facing dash cam
(Figure 22b), equipped with Wi-Fi, GPS, and a 1440p Quad HD (2K) camera (30 fps frame
rate), which can capture road signs and license plate numbers, so allowing drivers to save
photos and videos via a gesture sensor. Moreover, it also features a built-in G-Sensor that
automatically allows drivers to save road accident video for later review [104]. The
systems designed by Trakm8 also include the RH800 4G Mobile Digital Video Recorder
(MDVR), which supports up to four cameras and features telematics functionality,
including GPS tracking, driver behavior, and vehicle health monitoring (Figure 22c) [105].
The Trakm8 Insight telematics system can measure speed, over-revving, and harsh
acceleration and braking. Trakm8 Insight telematic utilizes GPS, fuel consumption,
driving trends, and dash cam data to offer information supporting management
processes.
Trakm8’s ConnectedCare vehicle health solution provides diagnostic data by
accessing the CANbus network of cars and commercial vehicles. This system identifies
Diagnostic Trouble Codes (DTCs) or engine fault codes characteristic of specific vehicle
issues and displays them on the Insight software platform. The Insight platform allows
remote monitoring of different vehicle warning lights, such as washer fluid level, ABS,
traction control, Diesel Particulate Filter (DPF), AdBlue levels, and tire pressure, and it
alerts vehicle owners and fleet managers to potential issues early for proactive
maintenance.
Trakm8’s Driver ID authentication software allows drivers to log into their assigned
vehicles using standard door entry cards, employee ID cards, or NFC-enabled mobile
devices. Drivers can utilize the ACC750 Driver ID, a combined driver identification and
feedback solution connected to different Trakm8 telematics solutions via Bluetooth or a
wired connection (Figure 22d) [106]. The ACC750 features a multi-function buon that
allows drivers to switch between business and private travel modes. Holding the buon
for three seconds triggers an alert sent to the paired telematics device, which then relays
the message to the server for monitoring through a web portal or mobile app.
(a) (b)
Sensors 2025, 25, x FOR PEER REVIEW 45 of 62
(c) (d)
Figure 22. Solution for vehicle condition monitoring by Trakm8: (a) 4G Integrated Telematics
Camera RH600 [103]; (b) RoadHawk DC-4 Dash Cam [104], (c) RH800 4G Mobile Digital Video
Recorder [105]; (d) ACC750 Driver ID and Feedback device [106].
Another leading IVMS available on the market is the TrackoBit GPS tracking
software proposed recently by InsightGeeks Solutions Pvt. Ltd (Noida, India), a mobility
technology company [107]. TrackoBit software is compatible with a lot of GPS hardware
and IoT devices, making it compatible with various manufacturers; its key features are
the driver behavior management and monitoring, ADAS interface, video telematics with
geospatial tracking solutions, and route planning and management. Driver behavior
management and monitoring is based on a GPS tracking system and various sensors,
providing data on driver performance, distinguishing between high- and under-
performing drivers, and verifying if they frequently deviate from the assigned route. This
software optimizes route planning to maximize operational efficiency and collects,
displays, and analyzes data related to sharp cornering, harsh acceleration and braking,
and over-speeding. The TrackoBit ADAS provides views of the vehicle from all angles
and is notified whenever the system recognizes any anomaly or dangerous situation
(Figure 23).
An AI-powered DMS and ADAS constitute the two pillars of Video Telematics by
TrackoBit system, which is integrated into vehicles via the OBD ports (Figure 24). Route
Planning and Management Software by TrackoBit offers trip planning and management
capabilities based on three steps (Figure 25). The first step, create tour, assists in designing
a personalized tour by entering the desired delivery location, seing the total trip time,
and arranging the designated stops. The second step, monitor route, assists in gaining
real-time visibility into vehicle locations and movements, monitoring their progress along
specified routes, and saving optimal paths for repeated journeys. The third step, manage
trips, assists in utilizing live tracking to oversee trips while vehicles are in transit.
Figure 22. Solution for vehicle condition monitoring by Trakm8: (a) 4G Integrated Telematics Camera
RH600 [
103
]; (b) RoadHawk DC-4 Dash Cam [
104
], (c) RH800 4G Mobile Digital Video Recorder [
105
];
(d) ACC750 Driver ID and Feedback device [106].
Trakm8’s ConnectedCare vehicle health solution provides diagnostic data by accessing
the CANbus network of cars and commercial vehicles. This system identifies Diagnostic
Trouble Codes (DTCs) or engine fault codes characteristic of specific vehicle issues and
displays them on the Insight software platform. The Insight platform allows remote
monitoring of different vehicle warning lights, such as washer fluid level, ABS, traction
control, Diesel Particulate Filter (DPF), AdBlue levels, and tire pressure, and it alerts vehicle
owners and fleet managers to potential issues early for proactive maintenance.
Trakm8’s Driver ID authentication software allows drivers to log into their assigned
vehicles using standard door entry cards, employee ID cards, or NFC-enabled mobile
devices. Drivers can utilize the ACC750 Driver ID, a combined driver identification and
feedback solution connected to different Trakm8 telematics solutions via Bluetooth or a
wired connection (Figure 22d) [
106
]. The ACC750 features a multi-function button that
allows drivers to switch between business and private travel modes. Holding the button
for three seconds triggers an alert sent to the paired telematics device, which then relays
the message to the server for monitoring through a web portal or mobile app.
Another leading IVMS available on the market is the TrackoBit GPS tracking software
proposed recently by InsightGeeks Solutions Pvt. Ltd (Noida, India), a mobility technology
company [
107
]. TrackoBit software is compatible with a lot of GPS hardware and IoT
devices, making it compatible with various manufacturers; its key features are the driver
behavior management and monitoring, ADAS interface, video telematics with geospatial
tracking solutions, and route planning and management. Driver behavior management
and monitoring is based on a GPS tracking system and various sensors, providing data
on driver performance, distinguishing between high- and under-performing drivers, and
Sensors 2025,25, 562 43 of 57
verifying if they frequently deviate from the assigned route. This software optimizes
route planning to maximize operational efficiency and collects, displays, and analyzes
data related to sharp cornering, harsh acceleration and braking, and over-speeding. The
TrackoBit ADAS provides views of the vehicle from all angles and is notified whenever the
system recognizes any anomaly or dangerous situation (Figure 23).
Sensors 2025, 25, x FOR PEER REVIEW 45 of 62
(c) (d)
Figure 22. Solution for vehicle condition monitoring by Trakm8: (a) 4G Integrated Telematics
Camera RH600 [103]; (b) RoadHawk DC-4 Dash Cam [104], (c) RH800 4G Mobile Digital Video
Recorder [105]; (d) ACC750 Driver ID and Feedback device [106].
Another leading IVMS available on the market is the TrackoBit GPS tracking
software proposed recently by InsightGeeks Solutions Pvt. Ltd (Noida, India), a mobility
technology company [107]. TrackoBit software is compatible with a lot of GPS hardware
and IoT devices, making it compatible with various manufacturers; its key features are
the driver behavior management and monitoring, ADAS interface, video telematics with
geospatial tracking solutions, and route planning and management. Driver behavior
management and monitoring is based on a GPS tracking system and various sensors,
providing data on driver performance, distinguishing between high- and under-
performing drivers, and verifying if they frequently deviate from the assigned route. This
software optimizes route planning to maximize operational efficiency and collects,
displays, and analyzes data related to sharp cornering, harsh acceleration and braking,
and over-speeding. The TrackoBit ADAS provides views of the vehicle from all angles
and is notified whenever the system recognizes any anomaly or dangerous situation
(Figure 23).
An AI-powered DMS and ADAS constitute the two pillars of Video Telematics by
TrackoBit system, which is integrated into vehicles via the OBD ports (Figure 24). Route
Planning and Management Software by TrackoBit offers trip planning and management
capabilities based on three steps (Figure 25). The first step, create tour, assists in designing
a personalized tour by entering the desired delivery location, seing the total trip time,
and arranging the designated stops. The second step, monitor route, assists in gaining
real-time visibility into vehicle locations and movements, monitoring their progress along
specified routes, and saving optimal paths for repeated journeys. The third step, manage
trips, assists in utilizing live tracking to oversee trips while vehicles are in transit.
Figure 23. ADAS functionalities by TrackoBit devices: forward, rear, or side collision alerts, signal
violation, lane switch alert, and over-speeding alert.
An AI-powered DMS and ADAS constitute the two pillars of Video Telematics by
TrackoBit system, which is integrated into vehicles via the OBD ports (Figure 24). Route
Planning and Management Software by TrackoBit offers trip planning and management
capabilities based on three steps (Figure 25). The first step, create tour, assists in designing
a personalized tour by entering the desired delivery location, setting the total trip time,
and arranging the designated stops. The second step, monitor route, assists in gaining
real-time visibility into vehicle locations and movements, monitoring their progress along
specified routes, and saving optimal paths for repeated journeys. The third step, manage
trips, assists in utilizing live tracking to oversee trips while vehicles are in transit.
Sensors 2025, 25, x FOR PEER REVIEW 46 of 62
Figure 23. ADAS functionalities by TrackoBit devices: forward, rear, or side collision alerts, signal
violation, lane switch alert, and over-speeding alert.
(a)
(b)
Figure 24. Video Telematics by TrackoBit system: DMS alert example (a), ADAS alert example (b).
(a) (b) (c)
Figure 25. Route planning and management software by TrackoBit: (a) create tour, (b) monitor
route, (c) manage trips.
Another interesting IVMS available on the market is the recently proposed
technological system by the automotive group FORVIA (Nanterre, France), an automotive
group that develops technologies to be installed inside or outside vehicles to support the
decision making of drivers and automated systems. The IVMS includes three
technologies: by-wire technology, radars and environmental sensors, and a vision system;
the first includes a brake pedal sensor (BPS) (Figure 26a), simulating the feeling of a
conventional brake system while electronically transmiing driver input. The BPS
transfers the driver’s action to the brake control unit, which provides a signal to the
braking system [108]. The by-wire technology also proposes a steering torque sensor; as
the driver turns the steering wheel, the sensor measures the torque applied to the steering
column. The sensor detects the torsion bar angle and provides information to the control
module, which calculates the necessary servo assistance to support the driver’s action
[109,110].
Figure 24. Video Telematics by TrackoBit system: DMS alert example (a), ADAS alert example (b).
Sensors 2025,25, 562 44 of 57
Sensors 2025, 25, x FOR PEER REVIEW 46 of 62
Figure 23. ADAS functionalities by TrackoBit devices: forward, rear, or side collision alerts, signal
violation, lane switch alert, and over-speeding alert.
(a)
(b)
Figure 24. Video Telematics by TrackoBit system: DMS alert example (a), ADAS alert example (b).
(a) (b) (c)
Figure 25. Route planning and management software by TrackoBit: (a) create tour, (b) monitor
route, (c) manage trips.
Another interesting IVMS available on the market is the recently proposed
technological system by the automotive group FORVIA (Nanterre, France), an automotive
group that develops technologies to be installed inside or outside vehicles to support the
decision making of drivers and automated systems. The IVMS includes three
technologies: by-wire technology, radars and environmental sensors, and a vision system;
the first includes a brake pedal sensor (BPS) (Figure 26a), simulating the feeling of a
conventional brake system while electronically transmiing driver input. The BPS
transfers the driver’s action to the brake control unit, which provides a signal to the
braking system [108]. The by-wire technology also proposes a steering torque sensor; as
the driver turns the steering wheel, the sensor measures the torque applied to the steering
column. The sensor detects the torsion bar angle and provides information to the control
module, which calculates the necessary servo assistance to support the driver’s action
[109,110].
Figure 25. Route planning and management software by TrackoBit: (a) create tour, (b) monitor route,
(c) manage trips.
Another interesting IVMS available on the market is the recently proposed technolog-
ical system by the automotive group FORVIA (Nanterre, France), an automotive group
that develops technologies to be installed inside or outside vehicles to support the decision
making of drivers and automated systems. The IVMS includes three technologies: by-wire
technology, radars and environmental sensors, and a vision system; the first includes
a brake pedal sensor (BPS) (Figure 26a), simulating the feeling of a conventional brake
system while electronically transmitting driver input. The BPS transfers the driver’s action
to the brake control unit, which provides a signal to the braking system [
108
]. The by-wire
technology also proposes a steering torque sensor; as the driver turns the steering wheel,
the sensor measures the torque applied to the steering column. The sensor detects the
torsion bar angle and provides information to the control module, which calculates the
necessary servo assistance to support the driver’s action [109,110].
Sensors 2025, 25, x FOR PEER REVIEW 47 of 62
(a) (b)
Figure 26. By-wire technology by FORVIA: brake-by-wire system (a) and steering torque sensor (b).
Radars and environment sensors by FORVIA include a 77GHZ radar, an eMirrors
camera, a road condition sensor, a rain-light sensor, and a parking system (manufactured
by FORVIA company, Nanterre, France) (Figure 27). The radar technology supports
ADASs, enhancing environmental detection by identifying stationary elements like road
boundaries and tracking dynamic objects such as pedestrians, cyclists, and vehicles [111].
The eMirrors camera replaces traditional external mirrors, providing a surrounding view;
the lightweight and aerodynamic design reduces CO
2
emissions and improves energy
efficiency. The road condition sensor detects water on the road in real time, estimating the
grip by vibration measurement through a piezoelectric sensor [112]. The rain–light sensor
is a multi-function device with the following capabilities: rain detection for automatic
wiper control, light sensing for automatic headlight activation and tunnel or overpass
detection, sun load measurement, head-up display brightness adjustment, and
humidity/temperature sensing to prevent windshield fogging [113].
Vision systems by FORVIA include an eMirror UX Safe engine and a surround view
system [114]. The eMirror UX Safe engine is a software that enhances mobility safety by
using a camera-based system to improve drivers’ visibility and situational awareness
(Figure 28a). It provides safety alerts, optimizes energy efficiency, and minimizes
distractions, ensuring clearer views even in challenging environments and lighting
conditions. The Vision system’s surround view enables a 360° panoramic perspective of
the car using a four-camera fisheye setup (Figure 28b). The surround view system has
been conceived to enhance safety, have beer visibility, and effortlessly maneuver.
Combining data from these cameras recreates the entire environment and offers features
such as virtual transparency with gaze detection.
(a) (b)
Figure 26. By-wire technology by FORVIA: brake-by-wire system (a) and steering torque sensor (b).
Radars and environment sensors by FORVIA include a 77GHZ radar, an eMirrors
camera, a road condition sensor, a rain-light sensor, and a parking system (manufactured by
FORVIA company, Nanterre, France) (Figure 27). The radar technology supports ADASs,
enhancing environmental detection by identifying stationary elements like road boundaries
and tracking dynamic objects such as pedestrians, cyclists, and vehicles [
111
]. The eMirrors
camera replaces traditional external mirrors, providing a surrounding view; the lightweight
and aerodynamic design reduces CO
2
emissions and improves energy efficiency. The road
condition sensor detects water on the road in real time, estimating the grip by vibration
measurement through a piezoelectric sensor [
112
]. The rain–light sensor is a multi-function
device with the following capabilities: rain detection for automatic wiper control, light
Sensors 2025,25, 562 45 of 57
sensing for automatic headlight activation and tunnel or overpass detection, sun load
measurement, head-up display brightness adjustment, and humidity/temperature sensing
to prevent windshield fogging [113].
Sensors 2025, 25, x FOR PEER REVIEW 47 of 62
(a) (b)
Figure 26. By-wire technology by FORVIA: brake-by-wire system (a) and steering torque sensor (b).
Radars and environment sensors by FORVIA include a 77GHZ radar, an eMirrors
camera, a road condition sensor, a rain-light sensor, and a parking system (manufactured
by FORVIA company, Nanterre, France) (Figure 27). The radar technology supports
ADASs, enhancing environmental detection by identifying stationary elements like road
boundaries and tracking dynamic objects such as pedestrians, cyclists, and vehicles [111].
The eMirrors camera replaces traditional external mirrors, providing a surrounding view;
the lightweight and aerodynamic design reduces CO
2
emissions and improves energy
efficiency. The road condition sensor detects water on the road in real time, estimating the
grip by vibration measurement through a piezoelectric sensor [112]. The rain–light sensor
is a multi-function device with the following capabilities: rain detection for automatic
wiper control, light sensing for automatic headlight activation and tunnel or overpass
detection, sun load measurement, head-up display brightness adjustment, and
humidity/temperature sensing to prevent windshield fogging [113].
Vision systems by FORVIA include an eMirror UX Safe engine and a surround view
system [114]. The eMirror UX Safe engine is a software that enhances mobility safety by
using a camera-based system to improve drivers’ visibility and situational awareness
(Figure 28a). It provides safety alerts, optimizes energy efficiency, and minimizes
distractions, ensuring clearer views even in challenging environments and lighting
conditions. The Vision system’s surround view enables a 360° panoramic perspective of
the car using a four-camera fisheye setup (Figure 28b). The surround view system has
been conceived to enhance safety, have beer visibility, and effortlessly maneuver.
Combining data from these cameras recreates the entire environment and offers features
such as virtual transparency with gaze detection.
(a) (b)
Sensors 2025, 25, x FOR PEER REVIEW 48 of 62
(c) (d)
Figure 27. Radars and environment sensors by FORVIA: 77GHz radar (a), e-Mirror camera (b),
SHAKE road condition sensor (c), HELLA Rain Light Sensor (d).
(a)
(b)
Figure 28. Vision systems by FORVIA: eMirror UX Safe engine (a), surround view system (b).
The last IVMS described in this section, although not yet on the market, is particularly
interesting for understanding future trends; it is a system developed by Social Self Driving
S.r.L (Como, Italy) that captures and replicates a driver’s unique driving style in
autonomous vehicles. This technology allows driverless cars to mimic the behavior of
their owners or professional drivers, selecting from a range of pre-recorded profiles [115].
The system utilizes hardware and software components commonly found in modern
vehicles with assisted or autonomous driving capabilities. It integrates various sensors,
including those that measure steering angles, torque, and speed, as well as sensors for
accelerator and brake pedal actions, yaw, pitch, roll, and both lateral and longitudinal
acceleration. From these data, the system profiles a driver’s style, which can be set or
shared through a dedicated cloud platform, allowing it to be programmed into
autonomous or semi-autonomous vehicles (Figure 29). The “Social Self Driving” app
allows users to access profiles made by others, creating a virtual marketplace where car
manufacturers, professional drivers, driving instructors, and public figures can offer and
sell customized driving programs [115].
(a)
(b)
Figure 27. Radars and environment sensors by FORVIA: 77GHz radar (a), e-Mirror camera (b),
SHAKE road condition sensor (c), HELLA Rain Light Sensor (d).
Vision systems by FORVIA include an eMirror UX Safe engine and a surround view
system [
114
]. The eMirror UX Safe engine is a software that enhances mobility safety by
using a camera-based system to improve drivers’ visibility and situational awareness (Fig-
ure 28a). It provides safety alerts, optimizes energy efficiency, and minimizes distractions,
ensuring clearer views even in challenging environments and lighting conditions. The
Vision system’s surround view enables a 360
◦
panoramic perspective of the car using a
four-camera fisheye setup (Figure 28b). The surround view system has been conceived to
enhance safety, have better visibility, and effortlessly maneuver. Combining data from these
cameras recreates the entire environment and offers features such as virtual transparency
with gaze detection.
Sensors 2025, 25, x FOR PEER REVIEW 48 of 62
(c) (d)
Figure 27. Radars and environment sensors by FORVIA: 77GHz radar (a), e-Mirror camera (b),
SHAKE road condition sensor (c), HELLA Rain Light Sensor (d).
(a)
(b)
Figure 28. Vision systems by FORVIA: eMirror UX Safe engine (a), surround view system (b).
The last IVMS described in this section, although not yet on the market, is particularly
interesting for understanding future trends; it is a system developed by Social Self Driving
S.r.L (Como, Italy) that captures and replicates a driver’s unique driving style in
autonomous vehicles. This technology allows driverless cars to mimic the behavior of
their owners or professional drivers, selecting from a range of pre-recorded profiles [115].
The system utilizes hardware and software components commonly found in modern
vehicles with assisted or autonomous driving capabilities. It integrates various sensors,
including those that measure steering angles, torque, and speed, as well as sensors for
accelerator and brake pedal actions, yaw, pitch, roll, and both lateral and longitudinal
acceleration. From these data, the system profiles a driver’s style, which can be set or
shared through a dedicated cloud platform, allowing it to be programmed into
autonomous or semi-autonomous vehicles (Figure 29). The “Social Self Driving” app
allows users to access profiles made by others, creating a virtual marketplace where car
manufacturers, professional drivers, driving instructors, and public figures can offer and
sell customized driving programs [115].
(a)
(b)
Figure 28. Vision systems by FORVIA: eMirror UX Safe engine (a), surround view system (b).
The last IVMS described in this section, although not yet on the market, is particu-
larly interesting for understanding future trends; it is a system developed by Social Self
Sensors 2025,25, 562 46 of 57
Driving S.r.L (Como, Italy) that captures and replicates a driver’s unique driving style in
autonomous vehicles. This technology allows driverless cars to mimic the behavior of their
owners or professional drivers, selecting from a range of pre-recorded profiles [
115
]. The
system utilizes hardware and software components commonly found in modern vehicles
with assisted or autonomous driving capabilities. It integrates various sensors, including
those that measure steering angles, torque, and speed, as well as sensors for accelerator and
brake pedal actions, yaw, pitch, roll, and both lateral and longitudinal acceleration. From
these data, the system profiles a driver’s style, which can be set or shared through a dedi-
cated cloud platform, allowing it to be programmed into autonomous or semi-autonomous
vehicles (Figure 29). The “Social Self Driving” app allows users to access profiles made
by others, creating a virtual marketplace where car manufacturers, professional drivers,
driving instructors, and public figures can offer and sell customized driving programs [
115
].
Sensors 2025, 25, x FOR PEER REVIEW 48 of 62
(c) (d)
Figure 27. Radars and environment sensors by FORVIA: 77GHz radar (a), e-Mirror camera (b),
SHAKE road condition sensor (c), HELLA Rain Light Sensor (d).
(a)
(b)
Figure 28. Vision systems by FORVIA: eMirror UX Safe engine (a), surround view system (b).
The last IVMS described in this section, although not yet on the market, is particularly
interesting for understanding future trends; it is a system developed by Social Self Driving
S.r.L (Como, Italy) that captures and replicates a driver’s unique driving style in
autonomous vehicles. This technology allows driverless cars to mimic the behavior of
their owners or professional drivers, selecting from a range of pre-recorded profiles [115].
The system utilizes hardware and software components commonly found in modern
vehicles with assisted or autonomous driving capabilities. It integrates various sensors,
including those that measure steering angles, torque, and speed, as well as sensors for
accelerator and brake pedal actions, yaw, pitch, roll, and both lateral and longitudinal
acceleration. From these data, the system profiles a driver’s style, which can be set or
shared through a dedicated cloud platform, allowing it to be programmed into
autonomous or semi-autonomous vehicles (Figure 29). The “Social Self Driving” app
allows users to access profiles made by others, creating a virtual marketplace where car
manufacturers, professional drivers, driving instructors, and public figures can offer and
sell customized driving programs [115].
(a)
(b)
Figure 29. (a) The vehicle learns the driving style from the driver, to be integrated with preloaded programs
and duplicated when same conditions occur. (b) Sharing data with other users via web platforms.
Table 12 compares the IVMSs analyzed in Section 5.2, considering the applications,
integrated electronic devices and sensors, monitored vehicular parameters, available con-
nectivity technologies and interfaces, and developed software to offer a comprehensive
evaluation of their capabilities.
Table 12. Comparison of commercial IVMSs illustrated in Section 5.2.
Commercial
IVMS Applications Cameras and
Other Devices Sensors
Monitored
Vehicle
Parameters
Connectivity
Technologies and
Interfaces
Software
Trakm8 system
[104–106]
Driver ID and
behavior, vehicle
health, ADAS
functionalities
4G camera, dash
cam, 4G mobile
digital video
recorder, driver
ID and
feedback device
G-Sensor
and radar
Washer fluid
level, ABS,
traction control,
DPF, AdBlue
levels, tire
pressure
BLE 4.2, GPS,
Wi-Fi, 4G
technology, CAN
1 and Tacho CAN
2 interfaces
ConnectedCare,
insight telematics,
driver ID
authentication
TrackoBit by
InsightGeeks
Solutions Pvt.
Ltd. [107]
Driver behavior,
ADAS
functionalities,
geospatial
tracking, route
planning and
management
AI-powered DMS
and camera Radar - GPS, 4G
technology
GPS tracking
software, video
telematics, route
planning and
management
software
IVMS by Forvia
[108–114]
By-wire
technology and
ADAS
functionalities
eMirrors and
surround-view
cameras
Brake and
steering torque
sensors, road
condition
piezoelectric
sensor, rain light
and parking
sensors, 77 GHz
radar
-OBD port, GPS,
4G technology eMirror UX Safe
engine
Sensors 2025,25, 562 47 of 57
5.3. Hardware and Software Commercial Solutions for Environmental and Traffic Monitoring
Traffic monitoring and management services can significantly enhance urban mobility
by reducing accidents and emissions. These services, based on seamless communication
between infrastructure, vehicles, and drivers, suggest alternative routes (if necessary) via
road alerts or real-time in-car messages. Various companies have introduced specialized
hardware and software solutions for traffic monitoring and driver assistance.
ClearView Intelligence (Milton Keynes, UK), a company specialized in mobility solu-
tions and data analytics, provides hardware and software for traffic monitoring, mobility
management, and predictive analytics. The AI-driven Insight Count and Classify software
monitors road traffic, counting and categorizing vehicles at key points, such as intersections
or highway segments [
116
]. Figure 30 illustrates its user interface, representing a solution
for enhancing travel safety and efficiency. The operating principle of Insight Count and
Classify software is based on some key features:
•
Data collection: the software uses devices (such as cameras or sensors) in different
environmental conditions to collect information about passing vehicles and traffic.
•Visual analysis: the software analyzes images of vehicles to identify their characteris-
tics, such as type (car, truck, motorcycle, etc.), size, and speed.
•
Data processing: the data are processed through algorithms that allow vehicles to be
counted and classified in real time.
•
Reporting and visualization: the analysis results are presented in reports and dash-
boards, providing information on traffic patterns, congestion peaks, and other metrics
useful for traffic planning and management.
Sensors 2025, 25, x FOR PEER REVIEW 50 of 62
intersections or highway segments [116]. Figure 30 illustrates its user interface,
representing a solution for enhancing travel safety and efficiency. The operating principle
of Insight Count and Classify software is based on some key features:
• Data collection: the software uses devices (such as cameras or sensors) in different
environmental conditions to collect information about passing vehicles and traffic.
• Visual analysis: the software analyzes images of vehicles to identify their
characteristics, such as type (car, truck, motorcycle, etc.), size, and speed.
• Data processing: the data are processed through algorithms that allow vehicles to be
counted and classified in real time.
• Reporting and visualization: the analysis results are presented in reports and
dashboards, providing information on traffic paerns, congestion peaks, and other
metrics useful for traffic planning and management.
Figure 30. Insight Count and Classify software interface developed by Clearview Intelligence
[116].
The Insight Count and Classify software targets stakeholders, including transport
and urban authorities, infrastructure project developers and mobility service providers,
to deal with traffic management, develop road infrastructure, and optimize vehicular
services. Clearview Intelligence also offers hardware solutions for traffic management.
Connex Traffic is a technology designed to tackle traffic management and road safety
through data utilization and analytics [117]. It employs a network of connected sensors
and devices to collect real-time traffic data (including vehicle flows, speeds, and road
conditions) from cameras, induction sensors, and GPS devices (Figure 31). These data are
processed using algorithms and AI techniques to identify traffic paerns, detect
congestion, and assess driving behavior. The analyzed information helps optimize traffic
signals, improve road signage, and develop strategies to reduce congestion. Real-time
updates can also be provided to drivers via apps or information panels.
Figure 30. Insight Count and Classify software interface developed by Clearview Intelligence [
116
].
The Insight Count and Classify software targets stakeholders, including transport
and urban authorities, infrastructure project developers and mobility service providers,
to deal with traffic management, develop road infrastructure, and optimize vehicular
services. Clearview Intelligence also offers hardware solutions for traffic management.
Connex Traffic is a technology designed to tackle traffic management and road safety
through data utilization and analytics [
117
]. It employs a network of connected sensors and
Sensors 2025,25, 562 48 of 57
devices to collect real-time traffic data (including vehicle flows, speeds, and road conditions)
from cameras, induction sensors, and GPS devices (Figure 31). These data are processed
using algorithms and AI techniques to identify traffic patterns, detect congestion, and
assess driving behavior. The analyzed information helps optimize traffic signals, improve
road signage, and develop strategies to reduce congestion. Real-time updates can also be
provided to drivers via apps or information panels.
Sensors 2025, 25, x FOR PEER REVIEW 51 of 62
Figure 31. Connex Traffic device is installed on the road: a typical scenario [117].
The M100 system by Clearview Intelligence is designed to be concerned with the
management and efficiency of transportation and urban mobility infrastructure (Figure
32). It includes sensors, cameras, radar, and other IoT devices to gather real-time data on
traffic flow, road user behavior, and environmental conditions [118]. These data are
processed using AI and ML algorithms to detect paerns, identify accidents, and monitor
congestion. By combining historical and real-time data, the system can predict traffic flow
and suggest strategies to optimize traffic and minimize waiting times. The M100 system
can also be integrated with other infrastructure and traffic management systems,
facilitating communication and coordination across different entities.
TomTom (Amsterdam, the Netherlands), a company specialized in navigation
software and applications, is known for its GPS navigation products and mapping
solutions. One of its latest solutions is TomTom Orbis Maps [119]. This mapping software
delivers maps and navigation data customized for various devices and applications. Orbis
Maps (Figure 33) integrates driver data with information from sources like Overture Maps
Foundation™ and OpenStreetMap, ensuring data acquisition at multiple levels. The
software combines geographic data from different sources, including satellite imagery,
road surveys, and local insights, to produce accurate and detailed maps. By combining
these mapping data with GPS technology, TomTom aims to provide efficient and reliable
navigation for users.
Figure 31. Connex Traffic device is installed on the road: a typical scenario [117].
The M100 system by Clearview Intelligence is designed to be concerned with the
management and efficiency of transportation and urban mobility infrastructure (Figure 32).
It includes sensors, cameras, radar, and other IoT devices to gather real-time data on traffic
flow, road user behavior, and environmental conditions [
118
]. These data are processed
using AI and ML algorithms to detect patterns, identify accidents, and monitor conges-
tion. By combining historical and real-time data, the system can predict traffic flow and
suggest strategies to optimize traffic and minimize waiting times. The M100 system can
also be integrated with other infrastructure and traffic management systems, facilitating
communication and coordination across different entities.
TomTom (Amsterdam, the Netherlands), a company specialized in navigation software
and applications, is known for its GPS navigation products and mapping solutions. One of
its latest solutions is TomTom Orbis Maps [
119
]. This mapping software delivers maps and
navigation data customized for various devices and applications. Orbis Maps (Figure 33)
integrates driver data with information from sources like Overture Maps Foundation™
and OpenStreetMap, ensuring data acquisition at multiple levels. The software combines
geographic data from different sources, including satellite imagery, road surveys, and local
insights, to produce accurate and detailed maps. By combining these mapping data with
GPS technology, TomTom aims to provide efficient and reliable navigation for users.
TomTom offers map updates that incorporate road changes, new constructions, and
real-time traffic data. Applications built on Orbis Maps feature an intuitive user interface,
making it easy to search for destinations, plan routes, and access information about live
traffic conditions and points of interest. Beyond standard road navigation, Orbis Maps
offers functionalities for various needs, including commercial vehicle route planning,
cycling, and pedestrian navigation, with data fitted to the user. Some TomTom services are
cloud-based, enabling access to maps and data across multiple devices. TomTom Orbis
Maps provides geolocation tools for companies involved in fleet management, delivery
route optimization, and other location-dependent operations.
Sensors 2025,25, 562 49 of 57
Sensors 2025, 25, x FOR PEER REVIEW 52 of 62
Figure 32. M100 system developed by Clearview Intelligence (a), application scenarios (b,c) [118].
(a)
(b)
(c)
Figure 32. M100 system developed by Clearview Intelligence (a), application scenarios (b,c) [118].
Sensors 2025, 25, x FOR PEER REVIEW 53 of 62
Figure 33. Different layers of TomTom’s Orbi Maps software.
TomTom offers map updates that incorporate road changes, new constructions, and
real-time traffic data. Applications built on Orbis Maps feature an intuitive user interface,
making it easy to search for destinations, plan routes, and access information about live
traffic conditions and points of interest. Beyond standard road navigation, Orbis Maps
offers functionalities for various needs, including commercial vehicle route planning,
cycling, and pedestrian navigation, with data fied to the user. Some TomTom services
are cloud-based, enabling access to maps and data across multiple devices. TomTom
Orbis Maps provides geolocation tools for companies involved in fleet management,
delivery route optimization, and other location-dependent operations.
The main differences that emerge between the Insight Count and Classify software
and Orbis Maps are the following:
• Intended use: the Insight Count and Classify software focuses on data collection and
analysis for traffic monitoring and management, whereas TomTom Orbis Maps
provides detailed digital maps for a variety of applications, including navigation and
vehicle automation;
• Technology: Insight Count and Classify uses physical sensors to collect data in the
field, whilst Orbis Maps combines data from various sources, including open-source
information and vehicle observations to create up-to-date digital maps.
• Target users: the Insight Count and Classify software is faced to road authorities and
city planners who need accurate traffic data, whereas Orbis Maps serves a broader
range of industries, including automotive and logistics, that require advanced digital
maps for a variety of applications.
Figure 33. Different layers of TomTom’s Orbi Maps software.
Sensors 2025,25, 562 50 of 57
The main differences that emerge between the Insight Count and Classify software
and Orbis Maps are the following:
•
Intended use: the Insight Count and Classify software focuses on data collection
and analysis for traffic monitoring and management, whereas TomTom Orbis Maps
provides detailed digital maps for a variety of applications, including navigation and
vehicle automation;
•
Technology: Insight Count and Classify uses physical sensors to collect data in the
field, whilst Orbis Maps combines data from various sources, including open-source
information and vehicle observations to create up-to-date digital maps.
•
Target users: the Insight Count and Classify software is faced to road authorities and
city planners who need accurate traffic data, whereas Orbis Maps serves a broader
range of industries, including automotive and logistics, that require advanced digital
maps for a variety of applications.
In summary, both software operate in the traffic management and mapping domain;
Insight Count and Classify is specialized in traffic data collection and analysis, whereas
Orbis Maps is a flexible and detailed digital mapping platform for industrial applications.
6. Challenges and Future Perspectives
The growing number of vehicles on the road have led to a significant increase in
accidents and frequent traffic congestion, often bringing cities to a standstill. This challenge
has driven research toward developing innovative solutions to improve road traffic through
driver assistance architectures. The goal is to reduce accidents by creating advanced driver
and vehicle monitoring systems while developing a communication network between
vehicles and infrastructure. Moreover, these innovative systems aim to optimize traffic flow
by offering drivers alternative routes to their destinations. Concerning driver monitoring,
low-intrusive systems, such as sensors integrated into the vehicle interior and cameras, are
identified as the most effective solutions. These systems maintain high driving comfort
while accurately detecting biophysical parameters to identify health anomalies, drowsiness,
stress, and intoxication. Similarly, the use of cameras for driver monitoring produces
excellent results but raises issues regarding data security and privacy, since sensitive data
are collected such as facial expressions, biometric details, and eye movements; therefore,
these non-intrusive solutions pave the way for the implementation of more advanced
data security techniques to prevent unauthorized access and tampering, according to the
main regulations (GPDR and CCPA). These challenges can be addressed by employing
end-to-end encryption techniques, localized processing, and data anonymization or pseudo-
anonymization techniques. Also, network technologies present significant challenges, the
main one being the latency reduction that allows them to provide real-time feedback and
overcome limitations related to limited bandwidth and congested communication channels.
The solutions to these problems lie in the use of advanced communication systems (5G-6G),
and where possible, in the exploitation of edge computing technologies. Another aspect is
the economic one, related to the investment costs for the acquisition of such technology, its
maintenance, and upgrading. Scalable designs, economies of scale, and possible govern-
ment incentives could allow wider adoption of these systems in next-generation vehicles.
Another issue concerns the precision and reliability of the proposed systems in detecting
the driver and vehicle parameters; since they are continuously subjected to vibrations and
mechanical stresses, their accuracy could be compromised. For this purpose, solutions
based on advanced AI models, advanced sensor systems, and continuous learning could
guarantee high performance for the DMS.
In the field of vehicle condition monitoring and driver behavior detection, a recurring
approach in the reviewed literature considers the vehicle’s OBD-II system. By integrating
Sensors 2025,25, 562 51 of 57
OBD-II data with Machine Learning (ML) algorithms, these solutions enable comprehen-
sive monitoring and predictive maintenance, including vehicle stability, emissions levels,
tire health, and driving style and behavior analysis. Finally, Vehicle-to-Infrastructure (V2I)
communication is highlighted as a critical component of traffic and road management. V2I
enables a continuous exchange of data useful for optimizing road signage, providing alter-
native routes, and allowing constant road condition monitoring. Emerging technologies
such as 6G communications and blockchain promise to significantly improve the efficiency,
safety, and reliability of Vehicle to Everything (V2X) systems. In particular, 6G technology
enables low latency, high reliability and data rate, and advanced connectivity, allowing
better traffic management and efficient vehicle coordination. Blockchain technology offers
decentralized and tamper-resistant data management, addressing several challenges in V2X
systems. It enables data security and integrity, availability of a decentralized public key
infrastructure (DPKI), and transparent and immutable ledgers. In November 2024, the U.S.
Federal Communications Commission approved new rules to promote the use of cellular
V2X (C-V2X) communication technologies. This initiative aims to improve road safety by
enabling vehicles to communicate with other vehicles, infrastructure, cyclists, and pedes-
trians, thereby facilitating warnings of hazardous conditions such as speeding vehicles,
weather conditions, or traffic congestion. Such capabilities enable targeted maintenance
efforts, ensuring greater safety and comfort for road users.
Future efforts should focus on advancing AI applications for driver monitoring and
the analysis of safety-critical events. A key development is integrating DMS data with road
scene analysis using AI techniques to enable early real-time crash prediction. A mobility
management system (MMS) can be developed based on intelligent vehicles able to monitor
vehicle and driver health, creating an interconnected network by exchanging information
in real time with on-the-road infrastructures, and informing users about traffic, crashes,
and environmental conditions. Each vehicle exchanges data via V2X technology with
surrounding ones and road infrastructure to integrate information about the driver’s status
with that provided by nearby vehicles and the road infrastructure. The multimodal analysis
of such information, based on DL algorithms, could improve the determination of the
driver’s risk level, contextualizing it to the scenario in which the vehicle is traveling.
Moreover, several strategies should be pursued to promote greater transparency
regarding AI implementation in DMS and risk events analysis. These should include
establishing industry standards, fostering research collaboration, and developing and
implementing open-source platforms. Driver Monitoring Systems (DMSs) are becoming a
key feature in modern vehicles, especially as we move toward advanced driver assistance
systems (ADASs) and even autonomous driving. To ensure these systems work effectively
across different vehicles and regions, standardization is crucial. Organizations like UNECE,
Euro NCAP, and NHTSA are laying the groundwork for what a DMS should do; in detail,
a standardized DMS should monitor and detect driver attention and health, fatigue, and
driving style. For example, from July 2024, the European Union has made it compulsory to
install alcohol detection systems in new vehicles, which communicate with the electronic
control unit to prevent the engine from starting when the driver is under the influence of
alcohol. One of the big challenges is ensuring that all the different pieces—cameras, sensors,
and software—can work across brands and models. Communication protocol to support
the data exchange between vehicles and infrastructures needs standardization to ensure
interoperability between different manufacturers, security, and ethical management of
collected data. Actually, standards like ISO 26262 [
120
], ISO 21448 [
121
], and General Data
Protection Regulation (GDPR) [
122
] ensure that electronic systems and privacy aspects are
safe and reliable.
Sensors 2025,25, 562 52 of 57
7. Conclusions
Automotive security systems and traffic monitoring technologies are essential to
keep cars safe, driving easier, and traffic flowing more smoothly by using advanced
surveillance and data-driven solutions. To gain deeper insights into the current systems
and future applications of AI tools in this field, this paper serves as a supplementary effort
to the knowledge of AI-driven solutions from both academic and industry perspectives.
Following a rigorous selection of scientific literature using the PRISMA methodology, this
paper provides a comprehensive overview of IoV solutions for driver state monitoring and
detection, vehicle monitoring, and traffic and road management. For each discussed topic,
a comparative analysis and related discussion are reported to highlight the strengths and
weaknesses of the analyzed solutions. Finally, the challenges and future perspectives are
presented, bringing out the main requirements and insights for the development of the
next generation of automotive monitoring systems.
Author Contributions: Conceptualization, P.V., G.R. and R.D.F.; methodology, P.V.; R.V., D.C. and
R.D.F.; validation, R.V., D.C. and R.D.F.; formal analysis, G.R., C.D.-V.-S., D.C. and R.D.F.; investiga-
tion, P.V., G.R., D.C. and R.D.F.; resources, P.V., G.R., R.V., D.C. and R.D.F.; data curation, G.R. and
R.D.F.; writing—original draft preparation, P.V., G.R. and R.D.F.; writing—review and editing, P.V.,
C.D.-V.-S., D.C. and R.V.; visualization, G.R., C.D.-V.-S. and R.D.F.; supervision, P.V., D.C. and R.V.;
funding acquisition, P.V. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: The data are available upon request.
Conflicts of Interest: The authors declare no conflicts of interest.
References
1.
Alalwany, E.; Mahgoub, I. Security and Trust Management in the Internet of Vehicles (IoV): Challenges and Machine Learning
Solutions. Sensors 2024,24, 368. [CrossRef] [PubMed]
2.
Bouhsissin, S.; Sael, N.; Benabbou, F. Driver Behavior Classification: A Systematic Literature Review. IEEE Access 2023,11,
14128–14153. [CrossRef]
3.
Guerrero-Ibáñez, J.; Zeadally, S.; Contreras-Castillo, J. Sensor Technologies for Intelligent Transportation Systems. Sensors 2018,
18, 1212. [CrossRef] [PubMed]
4.
Pandharipande, A.; Cheng, C.-H.; Dauwels, J.; Gurbuz, S.Z.; Ibanez-Guzman, J.; Li, G.; Piazzoni, A.; Wang, P.; Santra, A. Sensing
and Machine Learning for Automotive Perception: A Review. IEEE Sens. J. 2023,23, 11097–11115. [CrossRef]
5.
Mehta, A.A.; Padaria, A.A.; Bavisi, D.J.; Ukani, V.; Thakkar, P.; Geddam, R.; Kotecha, K.; Abraham, A. Securing the Future: A
Comprehensive Review of Security Challenges and Solutions in Advanced Driver Assistance Systems. IEEE Access 2024,12,
643–678. [CrossRef]
6.
Sohail, A.; Cheema, M.A.; Ali, M.E.; Toosi, A.N.; Rakha, H.A. Data-Driven Approaches for Road Safety: A Comprehensive
Systematic Literature Review. Saf. Sci. 2023,158, 105949. [CrossRef]
7.
Alanazi, F. A Systematic Literature Review of Autonomous and Connected Vehicles in Traffic Management. Appl. Sci. 2023,
13, 1789. [CrossRef]
8.
Nahid, M.S.H.; Tila, T.Z.; Seecharan, T.S. Detecting Emotional Arousal and Aggressive Driving Using Neural Networks: A Pilot
Study Involving Young Drivers in Duluth. Sensors 2024,24, 7109. [CrossRef] [PubMed]
9.
Mohammed, K.; Abdelhafid, M.; Kamal, K.; Ismail, N.; Ilias, A. Intelligent Driver Monitoring System: An Internet of Things-Based
System for Tracking and Identifying the Driving Behavior. Comput. Stand. Interfaces 2023,84, 103704. [CrossRef]
10.
Du, Z.; Deng, M.; Lyu, N.; Wang, Y. A Review of Road Safety Evaluation Methods Based on Driving Behavior. J. Traffic Transp.
Eng. 2023,10, 743–761. [CrossRef]
11.
Kaur, K.; Verma, H.K. IoV-Health: An Intelligent Integrated Emergency Health Monitoring and Alert Generation System. Soft
Comput. 2023. [CrossRef]
Sensors 2025,25, 562 53 of 57
12.
Affanni, A.; Rinaldo, R. Design and Realization of a Wearable Necklace for the Assessment of Driver Well-Being Through Heart
Rate and Blood Oxygen Saturation Monitoring. In Proceedings of the 2024 IEEE International Workshop on Metrology for
Automotive (MetroAutomotive), Bologna, Italy, 26–28 June 2024; IEEE: Piscataway, NJ, USA, 2024; pp. 153–158. [CrossRef]
13.
Gong, Z.; Yang, X.; Song, R.; Han, X.; Ren, C.; Shi, H.; Niu, J.; Li, W. Heart Rate Estimation in Driver Monitoring System Using
Quality-Guided Spectrum Peak Screening. IEEE Trans. Instrum. Meas. 2024,73, 5008214. [CrossRef]
14.
Sukumar, N.; Sivashankar, B.; Sumathi, P.; Maurya, O.P. Physiological and Physical Sensors for Stress Level, Drowsiness Detection,
and Behaviour Analysis. IEEE Trans. Consum. Electron. 2024,70, 656–668. [CrossRef]
15.
Safarov, F.; Akhmedov, F.; Abdusalomov, A.B.; Nasimov, R.; Cho, Y.I. Real-Time Deep Learning-Based Drowsiness Detection:
Leveraging Computer-Vision and Eye-Blink Analyses for Enhanced Road Safety. Sensors 2023,23, 6459. [CrossRef] [PubMed]
16.
Visconti, P.; de Fazio, R.; Costantini, P.; Miccoli, S.; Cafagna, D. Innovative Complete Solution for Health Safety of Children
Unintentionally Forgotten in a Car: A Smart Arduino-Based System with User App for Remote Control. IET Sci. Meas. Technol.
2020,14, 665–675. [CrossRef]
17.
Visconti, P.; de Fazio, R.; Costantini, P.; Miccoli, S.; Cafagna, D. Arduino-Based Solution for In-CarAbandoned Infants’ Controlling
Remotely Managed by Smartphone Application. J. Commun. Softw. Syst. 2019,15, 89–100. [CrossRef]
18.
Yang, X.G. Framework of Electric Vehicle Fault Diagnosis System Based on Diagnostic Communication. Int. J. Eng. 2024,37,
1194–1207. [CrossRef]
19.
Spiliotis, A.; Giannopoulos, F.; Spandonidis, C.; Gkemou, M.; Kalfa, N. Integration and Field Evaluation of an IoV System for
Enhancing Road Safety. Appl. Sci. 2022,12, 12262. [CrossRef]
20.
Visconti, P.; de Fazio, R.; Velazquez, R.; Al-Naami, B.; Ghavifekr, A.A. Self-Powered WiFi-Connected Monitoring Stations
for Environmental Pollution App-Based Control in Urban and Industrial Areas. In Proceedings of the 2022 8th International
Conference on Control, Instrumentation and Automation (ICCIA), Tehran, Iran, 2–3 March 2022; IEEE: Piscataway, NJ, USA, 2022;
pp. 1–6. [CrossRef]
21.
Márton, Z.; Szalay, I.; Fodor, D. Convolutional Neural Network-Based Tire Pressure Monitoring System. IEEE Access 2023,11,
70317–70332. [CrossRef]
22.
Hu, J.; Huang, M.-C.; Yu, X.B. Deep Learning Based on Connected Vehicles for Icing Pavement Detection. AI Civ. Eng. 2023,2, 1.
[CrossRef]
23.
Zhang, C.; Cheng, H.; Wu, R.; Ren, B.; Zhu, Y.; Peng, N. Development of a Traffic Congestion Prediction and Emergency Lane
Development Strategy Based on Object Detection Algorithms. Sustainability 2024,16, 10232. [CrossRef]
24.
Tahir, M.N.; Leviäkangas, P.; Katz, M. Connected Vehicles: V2V and V2I Road Weather and Traffic Communication Using Cellular
Technologies. Sensors 2022,22, 1142. [CrossRef] [PubMed]
25.
Lilhore, U.K.; Imoize, A.L.; Li, C.-T.; Simaiya, S.; Pani, S.K.; Goyal, N.; Kumar, A.; Lee, C.-C. Design and Implementation of an ML
and IoT Based Adaptive Traffic-Management System for Smart Cities. Sensors 2022,22, 2908. [CrossRef] [PubMed]
26.
De Fazio, R.; De Giorgi, M.; Cafagna, D.; Del-Valle-Soto, C.; Visconti, P. Energy Harvesting Technologies and Devices from
Vehicular Transit and Natural Sources on Roads for a Sustainable Transport: State-of-the-Art Analysis and Commercial Solutions.
Energies 2023,16, 3016. [CrossRef]
27.
Rathee, M.; Baˇci´c, B.; Doborjeh, M. Automated Road Defect and Anomaly Detection for Traffic Safety: A Systematic Review.
Sensors 2023,23, 5656. [CrossRef] [PubMed]
28.
Khanmohamadi, M.; Guerrieri, M. Advanced Sensor Technologies in CAVs for Traditional and Smart Road Condition Monitoring:
A Review. Sustainability 2024,16, 8336. [CrossRef]
29.
Page, M.J.; McKenzie, J.E.; Bossuyt, P.M.; Boutron, I.; Hoffmann, T.C.; Mulrow, C.D.; Shamseer, L.; Tetzlaff, J.M.; Akl, E.A.;
Brennan, S.E.; et al. The PRISMA 2020 Statement: An Updated Guideline for Reporting Systematic Reviews. BMJ 2021,372, n71.
[CrossRef]
30.
Page, M.J.; Moher, D.; Bossuyt, P.M.; Boutron, I.; Hoffmann, T.C.; Mulrow, C.D.; Shamseer, L.; Tetzlaff, J.M.; Akl, E.A.; Brennan,
S.E.; et al. PRISMA 2020 Explanation and Elaboration: Updated Guidance and Exemplars for Reporting Systematic Reviews. BMJ
2021,372, n160. [CrossRef] [PubMed]
31.
Welch, K.C.; Harnett, C.; Lee, Y.-C. A Review on Measuring Affect with Practical Sensors to Monitor Driver Behavior. Safety 2019,
5, 72. [CrossRef]
32.
Xiang, G.; Ou, Y.; Li, J.; Wang, L.; Hu, Y.; Wang, X.; Wu, X.; Peng, Y. Recovering Physiological Signals from Facial Videos: Recent
Advances and Applications in Intelligent Vehicles. IEEE Trans. Intell. Veh. 2024, 1–24. [CrossRef]
33.
Argüello-Prada, E.J. The Mountaineer’s Method for Peak Detection in Photoplethysmographic Signals. Rev. Fac. Ing. Univ. Antioq.
2019,90, 42–50. [CrossRef]
34.
Jiao, Y.; Zhang, C.; Chen, X.; Fu, L.; Jiang, C.; Wen, C. Driver Fatigue Detection Using Measures of Heart Rate Variability and
Electrodermal Activity. IEEE Trans. Intell. Transp. Syst. 2024,25, 5510–5524. [CrossRef]
35.
Putilov, A.A.; Donskaya, O.G. Construction and Validation of the EEG Analogues of the Karolinska Sleepiness Scale Based on the
Karolinska Drowsiness Test. Clin. Neurophysiol. 2013,124, 1346–1352. [CrossRef] [PubMed]
Sensors 2025,25, 562 54 of 57
36.
Leicht, L.; Walter, M.; Mathissen, M.; Antink, C.H.; Teichmann, D.; Leonhardt, S. Unobtrusive Measurement of Physiological
Features Under Simulated and Real Driving Conditions. IEEE Trans. Intell. Transp. Syst. 2022,23, 4767–4777. [CrossRef]
37.
Campos-Ferreira, A.E.; Lozoya-Santos, J.D.J.; Tudon-Martinez, J.C.; Mendoza, R.A.R.; Vargas-Martínez, A.; Morales-Menendez,
R.; Lozano, D. Vehicle and Driver Monitoring System Using On-Board and Remote Sensors. Sensors 2023,23, 814. [CrossRef]
[PubMed]
38.
Škori´c, T. Stress Level Detection Based on the Capacitive Electrocardiogram Signals of Driving Subjects. Sensors 2023,23, 9158.
[CrossRef] [PubMed]
39.
Ebrahimian, S.; Nahvi, A.; Tashakori, M.; Salmanzadeh, H.; Mohseni, O.; Leppänen, T. Multi-Level Classification of Driver
Drowsiness by Simultaneous Analysis of ECG and Respiration Signals Using Deep Neural Networks. Int. J. Environ. Res. Public
Health 2022,19, 10736. [CrossRef] [PubMed]
40.
Alguindigue, J.; Singh, A.; Narayan, A.; Samuel, S. Biosignals Monitoring for Driver Drowsiness Detection Using Deep Neural
Networks. IEEE Access 2024,12, 93075–93086. [CrossRef]
41.
BaHammam, A.S.; Nashwan, S.; Hammad, O.; Sharif, M.M.; Pandi-Perumal, S.R. Objective Assessment of Drowsiness and
Reaction Time during Intermittent Ramadan Fasting in Young Men: A Case-Crossover Study. Behav. Brain Funct. 2013,9, 32.
[CrossRef]
42.
Amidei, A.; Poli, A.; Iadarola, G.; Tramarin, F.; Pavan, P.; Spinsante, S.; Rovati, L. Driver Drowsiness Detection Based on
Variation of Skin Conductance from Wearable Device. In Proceedings of the 2022 IEEE International Workshop on Metrology for
Automotive (MetroAutomotive), Modena, Italy, 4–6 July 2022; IEEE: Piscataway, NJ, USA, 2022; pp. 94–98. [CrossRef]
43.
Díaz-Santos, S.; Cigala-Álvarez, Ó.; Gonzalez-Sosa, E.; Caballero-Gil, P.; Caballero-Gil, C. Driver Identification and Detection of
Drowsiness While Driving. Appl. Sci. 2024,14, 2603. [CrossRef]
44. Khan, M.A.; Nawaz, T.; Khan, U.S.; Hamza, A.; Rashid, N. IoT-Based Non-Intrusive Automated Driver Drowsiness Monitoring
Framework for Logistics and Public Transport Applications to Enhance Road Safety. IEEE Access 2023,11, 14385–14397. [CrossRef]
45.
Kundinger, T.; Sofra, N.; Riener, A. Assessment of the Potential of Wrist-Worn Wearable Sensors for Driver Drowsiness Detection.
Sensors 2020,20, 1029. [CrossRef]
46.
Vyas, D.A.; Chaturvedi, M. DriverSense: A Multi-Modal Framework for Advanced Driver Assistance System. In Proceedings of
the 2024 16th International Conference on COMmunication Systems & NETworkS (COMSNETS), Bengaluru, India, 3–7 January
2024; IEEE: Piscataway, NJ, USA, 2024; pp. 252–257. [CrossRef]
47.
Kumar, T.S.; Raman, R.; Karthikeyan, M.; Rawandale, C.J.; Sasikala, S.; Murugan, S. Drunk Driving Detection and Automatic Car
Ignition Locking System. In Proceedings of the 2023 International Conference on Self Sustainable Artificial Intelligence Systems
(ICSSAS), Erode, India, 18–20 October 2023; IEEE: Piscataway, NJ, USA, 2023; pp. 1469–1474. [CrossRef]
48.
Liu, J.; Luo, Y.; Ge, L.; Zeng, W.; Rao, Z.; Xiao, X. An Intelligent Online Drunk Driving Detection System Based on Multi-Sensor
Fusion Technology. Sensors 2022,22, 8460. [CrossRef] [PubMed]
49.
Swarna, M.; Saranya, G.; Vanathi, A.; Ashwini, R.; Arunsundar, B. An improved driver safety system based on drunk and drive
analyzing architecture using internet of things assisted alcohol identification sensor. In Proceedings of the 2024 International
Conference on Advances in Computing, Communication & Applied Information, Chennai, India, 9–10 May 2024; IEEE: Piscataway,
NJ, USA, 2024; pp. 1–7. [CrossRef]
50.
Wang, F.; Bai, D.; Liu, Z.; Yao, Z.; Weng, X.; Xu, C.; Fan, K.; Zhao, Z.; Chang, Z. A Two-Step E-Nose System for Vehicle Drunk
Driving Rapid Detection. Appl. Sci. 2023,13, 3478. [CrossRef]
51.
Akanni, J.; Adewusi, B.; Isa, A.A.; Alao, R.A. An Automated Prototype System for Alcohol Level Detection in Drunk Drivers and
Vehicle Control. Arid Zone J. Eng. Technol. Environ. 2023,19, 121–128.
52.
Cho, Y.; Lee, W.; Sin, H.; Oh, S.; Choi, K.C.; Jun, J.-H. Non-Invasive Alcohol Concentration Measurement Using a Spectroscopic
Module: Outlook for the Development of a Drunk Driving Prevention System. Sensors 2024,24, 2252. [CrossRef] [PubMed]
53.
Zhang, M.; Liu, X.; Jiang, B.; Chen, H.; Zhang, Q.; Yin, X. Innovative PVDF Nanofiber Films for Intelligent Driver Status
Monitoring Systems. In Proceedings of the COTA International Conference of Transportation Professionals (CICP2024), Shenzhen,
China, 23–26 July 2024; pp. 226–235. [CrossRef]
54.
Adib, F.; Mao, H.; Kabelac, Z.; Katabi, D.; Miller, R.C. Smart Homes That Monitor Breathing and Heart Rate. In Proceedings of the
33rd Annual ACM Conference on Human Factors in Computing Systems, Seoul, Republic of Korea, 18–23 April 2015; Association
for Computing Machinery: New York, NY, USA; pp. 837–846. [CrossRef]
55.
Vasantharaj, A.; Nandhagopal, N.; Karuppusamy, S.A.; Subramaniam, K. An In-Tire-Pressure Monitoring SoC Using FBAR
Resonator-Based ZigBee Transceiver and Deep Learning Models. Microprocess. Microsyst. 2022,95, 104709. [CrossRef]
56.
Kim, J.; Choi, S.B. Machine Learning Based Drive Fault Diagnosis and Isolation Algorithm for 4WIS4WID Vehicle Using Residual
Visualization. IEEE Trans. Intell. Veh. 2024,14, 1–15. [CrossRef]
57.
Kannan, S.R.; Partheeban, P.; Ramesh, R.; Elamparithi, P.N.; Somasundaram, K.; Selvi, R. Machine Learning and Vehicle Fault
Diagnosis System with IoT Enabled Data. In Proceedings of the 2023 International Conference on Data Science, Agents & Artificial
Intelligence (ICDSAAI), Chennai, India, 21–23 December 2023; IEEE: Piscataway, NJ, USA, 2023; pp. 1–9. [CrossRef]
Sensors 2025,25, 562 55 of 57
58. Min, H.; Fang, Y.; Wu, X.; Lei, X.; Chen, S.; Teixeira, R.; Zhu, B.; Zhao, X.; Xu, Z. A Fault Diagnosis Framework for Autonomous
Vehicles with Sensor Self-Diagnosis. Expert Syst. Appl. 2023,224, 120002. [CrossRef]
59.
Josephinshermila, P.; Priya, S.S.; Malarvizhi, K.; Hegde, R.; Gokul Pran, S.; Veerasamy, B. Accident Detection Using Automotive
Smart Black-Box Based Monitoring System. Meas. Sens. 2023,27, 100721. [CrossRef]
60.
Fichtinger, A.; Edelmann, J.; Plochl, M.; Holl, M. Aquaplaning Detection Using Effect-Based Methods: An Approach Based on a
Minimal Set of Sensors, Electronic Stability Control, and Drive Torques. IEEE Veh. Technol. Mag. 2021,16, 20–28. [CrossRef]
61.
Gustafsson, F. Estimation and Change Detection of Tire-Road Friction Using the Wheel Slip. In Proceedings of the Joint Conference
on Control Applications Intelligent Control and Computer Aided Control System Design, Dearborn, MI, USA, 15–18 September
1996; IEEE: Piscataway, NJ, USA, 1996; pp. 99–104. [CrossRef]
62.
Lee, H.; Hwang, K.; Kang, M.; Song, J. Black Ice Detection Using CNN for the Prevention of Accidents in Automated Vehicle. In
Proceedings of the 2020 International Conference on Computational Science and Computational Intelligence (CSCI), Las Vegas,
NV, USA, 16–18 December 2020; IEEE: Piscataway, NJ, USA, 2020; pp. 1189–1192. [CrossRef]
63.
Bhadraray, S.; Dey, S.; Amith, T.S.; Ramaiah, S. Autonomous Vehicle System for Pothole Detection and Avoidance. In Proceedings
of the 2023 IEEE 8th International Conference for Convergence in Technology (I2CT), Lonavla, India, 7–9 April 2023; IEEE:
Piscataway, NJ, USA, 2023; pp. 1–6. [CrossRef]
64.
Raja, G.; Anbalagan, S.; Senthilkumar, S.; Dev, K.; Qureshi, N.M.F. SPAS: Smart Pothole-Avoidance Strategy for Autonomous
Vehicles. Trans. Intell. Transp. Syst. 2022,23, 19827–19836. [CrossRef]
65.
Tselentis, D.I.; Papadimitriou, E. Driver Profile and Driving Pattern Recognition for Road Safety Assessment: Main Challenges
and Future Directions. IEEE Open J. Intell. Transp. Syst. 2023,4, 83–100. [CrossRef]
66.
Nouh, R.; Singh, M.; Singh, D. SafeDrive: Hybrid Recommendation System Architecture for Early Safety Predication Using
Internet of Vehicles. Sensors 2021,21, 3893. [CrossRef] [PubMed]
67.
Shichkina, Y.; Fatkieva, R.; Kopylov, M. Analysis of Driving Style Using Self-Organizing Maps to Analyze Driver Behavior. Int. J.
Electr. Comput. Eng. 2024,14, 2212–2225. [CrossRef]
68.
Chhabra, R.; Singh, S.; Khullar, V. Privacy Enabled Driver Behavior Analysis in Heterogeneous IoV Using Federated Learning.
Eng. Appl. Artif. Intell. 2023,120, 105881. [CrossRef]
69.
Malik, M.; Nandal, R.; Dalal, S.; Jalglan, V.; Le, D.-N. Driving Pattern Profiling and Classification Using Deep Learning. Intell.
Autom. Soft Comput. 2021,28, 887–906. [CrossRef]
70.
Lattanzi, E.; Freschi, V. Machine Learning Techniques to Identify Unsafe Driving Behavior by Means of In-Vehicle Sensor Data.
Expert Syst. Appl. 2021,176, 114818. [CrossRef]
71.
Dong, B.-T.; Lin, H.-Y. An On-Board Monitoring System for Driving Fatigue and Distraction Detection. In Proceedings of the 2021
22nd IEEE International Conference on Industrial Technology (ICIT), Valencia, Spain, 10–12 March 2021; IEEE: Piscataway, NJ,
USA, 2021; Volume 1, pp. 850–855. [CrossRef]
72.
Shan, T.; Li, L.; Mu, H.; Yu, L.; Liu, Z. Tire Pressure Monitoring System Based on Resonant Frequency of Wheel Speed Spectrum.
In Proceedings of the 2022 IEEE 5th International Conference on Electronics Technology (ICET), Chengdu, China, 13–16 May
2022; IEEE: Piscataway, NJ, USA, 2022; pp. 623–628. [CrossRef]
73.
Huo, B.; Zhang, M.; Zhong, S.; Zhou, F. Design of Indirect Tire Pressure Monitoring System Based on Wheel Speed Signal. J. Phys.
Conf. Ser. 2023,2528, 012045. [CrossRef]
74.
Pomoni, M. Exploring Smart Tires as a Tool to Assist Safe Driving and Monitor Tire–Road Friction. Vehicles 2022,4, 744–765.
[CrossRef]
75.
Yu, H.; Zhong, Y.; Li, W. A Transformer Model Based Method for Detecting Tire Wear Degree. In Proceedings of the 2024 9th
International Conference on Intelligent Computing and Signal Processing (ICSP), Xian, China, 19–21 April 2024; IEEE: Piscataway,
NJ, USA, 2024; pp. 1618–1624. [CrossRef]
76.
Anusha, N.; Jeslin, J.G.; Srividhya, V.; Gupta, N.S.; Meenakshi, R.; Srinivasan, C. Cloud-Enabled Neural Networks for Intelligent
Vehicle Emissions Tracking and Analysis. In Proceedings of the 2024 International Conference on Automation and Computation
(AUTOCOM), Dehradun, India, 14–16 March 2024; IEEE: Piscataway, NJ, USA, 2024; pp. 232–236. [CrossRef]
77.
Seo, J.; Park, S. Optimizing Model Parameters of Artificial Neural Networks to Predict Vehicle Emissions. Atmos. Environ. 2023,
294, 119508. [CrossRef]
78.
Andrade, P.; Silva, I.; Silva, M.; Flores, T.; Cassiano, J.; Costa, D.G. A TinyML Soft-Sensor Approach for Low-Cost Detection and
Monitoring of Vehicular Emissions. Sensors 2022,22, 3838. [CrossRef] [PubMed]
79.
Oladimeji, D.; Gupta, K.; Kose, N.A.; Gundogan, K.; Ge, L.; Liang, F. Smart Transportation: An Overview of Technologies and
Applications. Sensors 2023,23, 3880. [CrossRef] [PubMed]
80.
Saleem, M.; Abbas, S.; Ghazal, T.M.; Adnan Khan, M.; Sahawneh, N.; Ahmad, M. Smart Cities: Fusion-Based Intelligent Traffic
Congestion Control System for Vehicular Networks Using Machine Learning Techniques. Egypt. Inform. J. 2022,23, 417–426.
[CrossRef]
Sensors 2025,25, 562 56 of 57
81.
Wu, Q.; Wang, W.; Fan, P.; Fan, Q.; Zhu, H.; Letaief, K.B. Cooperative Edge Caching Based on Elastic Federated and Multi-Agent
Deep Reinforcement Learning in Next-Generation Networks. IEEE Trans. Netw. Serv. Manag. 2024,21, 4179–4196. [CrossRef]
82.
Mandal, V.; Mussah, A.R.; Jin, P.; Adu-Gyamfi, Y. Artificial Intelligence-Enabled Traffic Monitoring System. Sustainability 2020,
12, 9177. [CrossRef]
83.
Humayun, M.; Afsar, S.; Almufareh, M.F.; Jhanjhi, N.Z.; AlSuwailem, M. Smart Traffic Management System for Metropolitan
Cities of Kingdom Using Cutting Edge Technologies. J. Adv. Transp. 2022,2022, 4687319. [CrossRef]
84.
Putra, A.S.; Warnars, H.L.H.S. Intelligent Traffic Monitoring System (ITMS) for Smart City Based on IoT Monitoring. In
Proceedings of the 2018 Indonesian Association for Pattern Recognition International Conference (INAPR), Jakarta, Indonesia,
7–8 September 2018; IEEE: Piscataway, NJ, USA, 2018; pp. 161–165. [CrossRef]
85.
Sarrab, M.; Pulparambil, S.; Awadalla, M. Development of an IoT Based Real-Time Traffic Monitoring System for City Governance.
Glob. Transit. 2020,2, 230–245. [CrossRef]
86.
Barbosa, R.; Ogobuchi, O.D.; Joy, O.O.; Saadi, M.; Rosa, R.L.; Al Otaibi, S.; Rodríguez, D.Z. IoT Based Real-Time Traffic Monitoring
System Using Images Sensors by Sparse Deep Learning Algorithm. Comput. Commun. 2023,210, 321–330. [CrossRef]
87.
Kheder, M.Q.; Mohammed, A.A. Real-Time Traffic Monitoring System Using IoT-Aided Robotics and Deep Learning Techniques.
Kuwait J. Sci. 2024,51, 100153. [CrossRef]
88.
Dhingra, S.; Madda, R.B.; Patan, R.; Jiao, P.; Barri, K.; Alavi, A.H. Internet of Things-Based Fog and Cloud Computing Technology
for Smart Traffic Monitoring. Internet Things 2021,14, 100175. [CrossRef]
89.
Khan, N.A.; Jhanjhi, N.Z.; Brohi, S.N.; Usmani, R.S.A.; Nayyar, A. Smart Traffic Monitoring System Using Unmanned Aerial
Vehicles (UAVs). Comput. Commun. 2020,157, 434–443. [CrossRef]
90.
Singh, S.; Singh, J.; Goyal, S.B.; Singh Sehra, S.; Ali, F.; Ayad Alkhafaji, M.; Singh, R. A Novel Framework to Avoid Traffic
Congestion and Air Pollution for Sustainable Development of Smart Cities. Sustain. Energy Technol. Assess. 2023,56, 103125.
[CrossRef]
91.
Åstrand, M.; Jakobsson, E.; Lindfors, M.; Svensson, J. A System for Underground Road Condition Monitoring. Int. J. Min. Sci.
Technol. 2020,30, 405–411. [CrossRef]
92.
Ye, Z.; Wei, Y.; Yang, S.; Li, P.; Yang, F.; Yang, B.; Wang, L. IoT-Enhanced Smart Road Infrastructure Systems for Comprehensive
Real-Time Monitoring. Internet Things Cyber-Phys. Syst. 2024,4, 235–249. [CrossRef]
93.
Abdelmalak, M.E.S.; Khodadadi, N.; Zaki, A.M.; Eid, M.M.; Rizk, F.H.; Ibrahim, A.; Abdelhamid, A.A.; Abualigah, L.; El-kenawy,
E.S.M. Pothole Detection in Asphalt Roads: A Comprehensive Approach for Enhanced Road Maintenance and Safety with
AlexNet Model. In Proceedings of the 2024 International Telecommunications Conference (ITC-Egypt), Cairo, Egypt, 22–25 July
2024; IEEE: Piscataway, NJ, USA, 2024; pp. 269–274. [CrossRef]
94.
Ye, Z.; Wei, Y.; Li, J.; Yan, G.; Wang, L. A Distributed Pavement Monitoring System Based on Internet of Things. J. Traffic Transp.
(Eng. Ed.) 2022,9, 305–317. [CrossRef]
95.
Gaspar, G.; Dudak, J.; Behulova, M.; Stremy, M.; Budjac, R.; Sedivy, S.; Tomas, B. IoT-Ready Temperature Probe for Smart
Monitoring of Forest Roads. Appl. Sci. 2022,12, 743. [CrossRef]
96.
Hajder, M.; Hajder, P.; Hajder, L.; Liput, M.; Kolbusz, J. Supporting Winter Road Maintenance Procedures with the Use of
Distributed Measurements Based on IoT, Thermodynamic Models and Machine Learning. In Proceedings of the 2023 IEEE
International Conference on Pervasive Computing & Communication Workshops, Atlanta, GA, USA, 13–17 March 2023; IEEE:
Piscataway, NJ, USA, 2023; pp. 515–520. [CrossRef]
97.
Chen, Z.; Xiong, G.; Sun, Y.; Li, Y.; Li, Y. An Internet-of-Things-Enabled System for Road Icing Detection and Prediction. IEEE
Internet Things J. 2022,9, 20257–20269. [CrossRef]
98.
Kotus, J.; Szwoch, G. Detection of Water on Road Surface with Acoustic Vector Sensor. Sensors 2023,23, 8878. [CrossRef] [PubMed]
99.
Smart Eye Pro 12. Available online: https://www.smarteye.se/news/unveiling-smart-eye-pro-12-redefining-eye-tracking-
technology/ (accessed on 10 October 2024).
100.
Advanced Cabin Sensing Solution. Available online: https://conti-engineering.com/services/components/cabin-sensing/
(accessed on 11 October 2024).
101.
News Release—Magna Lands Industry-First Award for Integrated Driver and Occupant Monitoring System. Available
online: https://www.magna.com/company/newsroom/releases-archive/release/2022/02/07/news-release---magna-lands-
industry-first-award-for-integrated-driver-and-occupant-monitoring-system (accessed on 11 October 2024).
102.
Driver Fatigue Monitoring System for Teen and Fleet Driver Safety. Available online: https://speedir.com/driver-alert-fatigue-
monitoring-system-for-fleet-safety-night-driving-student-teen-driver/ (accessed on 30 October 2024).
103.
RH600 4G Integrated Telematics Camera. Available online: https://www.trakm8.com/vehicle-cameras/rh600-integrated-
telematics-camera (accessed on 5 November 2024).
104.
RoadHawk DC-4 Dash Cams. Available online: https://www.trakm8.com/vehicle-cameras/dash-cams/ (accessed on 7 Novem-
ber 2024).
Sensors 2025,25, 562 57 of 57
105.
RH800 4G Mobile Digital Video Recorder. Available online: https://www.trakm8.com/wp-content/uploads/Trakm8_RH800
_Product_Information_Sheet.pdf (accessed on 11 November 2024).
106.
ACC750 Driver ID & Feedback. Available online: https://www.trakm8.com/wp-content/uploads/Driver_Feedback_and_
Driver_Display.pdf (accessed on 11 November 2024).
107.
Trackobit: Advanced-Driver-Assistance-Systems. Available online: https://trackobit.com/advanced-driver-assistance-systems
(accessed on 8 January 2025).
108.
Brake Pedal Sensor. Available online: https://www.hella.com/microsite-electronics/en/Products/Brake-Pedal-Sensor-4923/
(accessed on 13 November 2024).
109.
Road Wheel Actuator. Available online: https://www.hella.com/microsite-electronics/en/Products/Road-Wheel-Actuator-49
13/ (accessed on 14 November 2024).
110.
Steering Torque Only Sensor. Available online: https://www.hella.com/microsite-electronics/en/Products/Steering-Torque-
Only-Sensor-4921/ (accessed on 13 November 2024).
111.
Radar 77GHZ. Available online: https://www.hella.com/microsite-electronics/en/Products/Radar-77GHZ-4919/ (accessed on
14 November 2024).
112.
SHAKE—Road Condition Sensor. Available online: https://www.hella.com/microsite-electronics/en/Products/SHAKE-Road-
Condition-Sensor-4979/ (accessed on 15 November 2024).
113.
Rain Light Sensor. Available online: https://www.hella.com/microsite-electronics/en/Products/Rain-Light-Sensor-4973/
(accessed on 18 November 2024).
114.
Safe and Automated Driving. Available online: https://www.forvia.com/en/pioneering-technologies/safe-automated-driving-
0/vision-systems-wire-control (accessed on 18 November 2024).
115.
Social Self Driving. Available online: https://www.socialselfdriving.com/en/home-social-self-driving-en/ (accessed on
18 November 2024).
116.
Insight Count & Classify. Available online: https://www.clearview-intelligence.com/products/insight-count-classify/ (accessed
on 20 November 2024).
117.
Connex Traffic. Available online: https://www.clearview-intelligence.com/products/connex-traffic/ (accessed on 21 November
2024).
118.
M100 Wireless Vehicle Detection System. Available online: https://www.clearview-intelligence.com/products/m100-wireless-
vehicle-detection-system/ (accessed on 21 November 2024).
119.
TomTom Orbis Maps—Digital Mapping Solutions. Available online: https://www.tomtom.com/tomtom-orbis-maps/ (accessed
on 25 November 2024).
120.
ISO 26262-1:2018, Road Vehicles — Functional Safety. Available online: https://www.iso.org/standard/68383.html (accessed on
20 November 2024).
121.
ISO 21448:2022, Road Vehicles—Safety of the Intended Functionality. Available online: https://www.iso.org/standard/77490.
html (accessed on 20 November 2024).
122. General Data Protection Regulation -GDPR. Available online: https://gdpr-info.eu/ (accessed on 20 November 2024).
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.
