IoT Sensor Systems Used for Monitoring Road Infrastructure

IoT Sensor Systems Used for Monitoring Road Infrastructure: History

Please note this is an old version of this entry, which may differ significantly from the current revision.

Subjects: Computer Science, Interdisciplinary Applications

Contributor:

An intelligent transportation system is one of the fundamental goals of the smart city concept. The Internet of Things (IoT) concept is a basic instrument to digitalize and automatize the process in the intelligent transportation system. Digitalization via the IoT concept enables the automatic collection of data usable for management in the transportation system. The IoT concept includes a system of sensors, actuators, control units and computational distribution among the edge, fog and cloud layers.

edge computing
transportation tasks
motion detection
object tracking
object detection
object classification
intrusive sensors
non-intrusive sensors
IoT sensor systems
monitoring the road infrastructure

1. Introduction

Transportation systems are the backbone of every economy in the world. Internet of Things (IoT) concepts applied to road transportation infrastructure can create the part of an intelligent transportation system. An intelligent transportation system could increase traffic efficiency and safety because the road’s monitoring system would be reliable.

Organizing fluent and safe traffic is a priority in every country. By defining monitoring tasks in intelligent transportation systems, it is necessary to achieve this goal and characterize the data to be collected. In compliance with [1], there are measured data about safety, the diagnostics of a vehicle or infrastructure, traffic management, driver’s assistance, environment’s impact and user’s ability to drive. The sensor systems used for traffic management and safety data use the techniques for vehicle motion detection and object tracking. The traffic and safety management system requires monitoring tasks such as vehicle type classification and counting, automatic license plate recognition, incident detection, parking management and speed measurement. Regarding the opinion of the authors in the scientific article [2], many sensor systems have been developed to solve these tasks, but their effectiveness has not significantly improved over the past decade.

2. IoT Concepts

The IoT is a concept of connecting devices to the internet. This connection helps to create a digital copy of the physical world. The main attributes of the IoT concept are sensors, actuators, control units, and the internet connecting these entities. The Internet is an instrument that connects each part of the IoT concept, and actuators are devices used to execute actions in the IoT concept. In the opinion of the authors [3,4], the IoT concept used for intelligent transportation systems is an important step in improving the quality of the transportation industry.

3. Transportation Sensoric Systems

Most countries and cities try to optimize the traffic and transportation control via IoT concepts to develop smart cities and smart transportation concepts. A lot of different routes have implemented IoT concepts. Still, the development of new sensors, computational devices, and protocol concepts create new possibilities to improve the effectiveness of these systems including maintenance, financial and computational capacities [27,28]. The mentioned tasks in transportation have their specifications creating a constraint to install some kinds of sensors. One of the vital requirements of the sensor is the ability to produce actual data where the control unit considers these data. There could be created either a virtual sensor based on these physical sensors. Here, there is describe a survey of the most common sensors used in transportation and traffic systems and their taxonomy shown in Table 1 from the view of deployment.

Table 1. Taxonomy of the sensors used in transportation.

Intrusive Sensors	In Roadway Based Sensors	Piezoelectric Sensor
		Fiber Bragg Gratting Sensor
		Pneumatic Tube Detector
		Vibration Sensor
		Loop Detector
		Magnetic Sensor
Non-intrusive Sensors Sensors	Side Roadway Based Sensors	RFID
		Acoustic Sensor
		Laser Scanner
		Radar
		Wi-Fi
	Over Roadway Based Sensors	Lidar
		Passive Infrared Sensor
		Ultrasonic Sensor
		Camera Systems

In real-time data analysis, intelligent transportation systems use sensors that measure humidity, temperature, wind speed, precipitation, and visibility. They play a critical role in forecasting weather conditions that impact traffic safety. However, this paper does not include the sensors used for the weather forecast in the taxonomy of sensors mentioned in Table 1 because they do not solve tasks based on motion detection and object tracking.

4. Transportation Tasks Based on Motion Detection and Object Tracking

The monitoring tasks are focused on the object’s motion detection and object tracking. Automatized monitoring tasks via IoT concepts in transportation could improve the conditions of traffic situations.

In-vehicle sensors could detect or warn the driver before a collision with other objects [58]. Development and research in artificial intelligence (AI) applications enable the building of driverless vehicles that consider data collected from sensors in real-time and process these to control vehicles while maintaining security standards [59].

4.1. High-Occupancy Vehicle Lane Management

Several countries try to reduce traffic congestion by motivating citizens to share cars or use public transportation. For this purpose, high-occupancy vehicle (HOV) lanes have been built, which serve only vehicles with two or more passengers. These HOV lanes enable drivers to reach their target destination quickly, avoiding potential traffic congestion [60]. HOV lane management systems are proposed to check the count of passengers in every vehicle. The system sorts the vehicles that have permission to use the HOV lanes. There are two options for checking the passenger count in the vehicles. The first one is using sensors for person detection from the checking station [61]. The second one is using seat occupancy sensors installed on the car’s interior [62]. The information extracted from the sensors could be sent to the checking station via the dedicated communication protocol.

4.2. Incident Detection

We detect incidents via various methods and different systems. Transportation incidents could happen in various scenarios, such as a vehicle fire inside a tunnel [63], traffic jams resulting from a damaged car [64], or chain car accidents [65]. For this purpose, many different types of sensors and communication devices are used to contact emergency services. We can divide incident identification into two aspects. The identification of incidents via infrastructure—for example, the camera surveillance system in a tunnel identifies a fire in a tunnel, which alarms the operator to begin the rescue of people inside the tunnel [66]. The second possibility is using in-vehicle sensors. For example, the ultrasonic sensors detect the collision, and the car’s board computer contacts the emergency service by wireless internet connection [67].

4.3. Vehicle Counting

Many paid highways use a counting vehicle system to monitor usage [68]. The counting vehicle system produces information about the frequency of visiting the routes. This information is used in future planning for maintenance or building new highways or roads [69,70]. The advanced counting vehicle system categorizes vehicles into many categories, such as buses, trucks, or cars. These collected data could help manage traffic.

4.4. License Plate Recognition

License plate recognition is an essential task in road surveillance. This task can identify cars with detected anomalous behavior. We can usually apply the methods of automatic license plate reading after finding the car’s position via some object detector. According to [71], the automatic license plate recognition (ALPR), process is divided into three steps:

Localize the position of the license plate in car.
Segment characters from the background.
Apply optical character recognition (OCR) methods.

This task is essential for checking the priced parking places [72] or identification of stolen cars [73].

4.5. Vehicle Type Classification

Many roads have some restrictions on the types of vehicles they allow because of safety, flow, and maintenance costs. Trucks, due to their heavy weight, are prohibited from using some routes [74]. In many countries, truck drivers have to pay extra charges to use highways due to higher maintenance costs [75]. Sometimes, the large size of trucks or buses limits passing through some sections without a traffic incident [76]. The data collected from the categorized count of vehicles by particular sections of the roads can improve the management of traffic lights to avoid traffic jams [66,77]. This information can be used by the dispatchers of public transport to react to special events.

4.6. Speed Measurement

Safety is considered a target priority in transportation. High-speed driving vehicles increase the risk of an accident. The increased kinetic energy of high-speed driving impacts the crash consequences after the collision with another vehicle or object. The risk of accidents in high-speed driving is related to limited human reaction time. States worldwide use speed limit restrictions in traffic management to reduce accident risks [78]. One of the aims of an intelligent transportation system is to check adherence to speed limit restrictions. The police use sensor systems that detect vehicle motion and track vector movements for speed measurement [79].

4.7. Parking Management

The increased number of people buying the cars influence the fact that they are widely used as a kind of individual transport method. Many people ride to work, school, or shop by car, occupying parking slots across a city or town. Free parking slots are a limited resource in many cities; therefore, many parking houses or places are priced [80]. Priced parking places require some automatic system of checking all the car owners to pay for parking slots [72]. A lot of automatic parking place checking is based on a sensor system similar to the vehicle counting system.

4.8. Weight Measurement

Worldwide, there are many specialized sections of roads that have specific vehicle weight constraints. These constraints exist because of road maintenance or security reasons [74]. To avoid the entry of overweight vehicles, sensors are installed that measure the vehicle’s weight, and the system can close the entry gate. For example, bridges are places where a vehicle’s axle weight restriction is necessary to keep traffic safe [81].

5. Applications the Sensor System in Transportation Tasks

Table 2 represents state of the art in sensor systems used for intelligent monitoring transportation tasks in the last decade. The literature review in Table 2 refers to identification of research challenge for each sensor system in tasks mentioned in the table.

Table 2. Table of the transportation tasks solved via the sensors.

	HOV Lane Management	Incident Detection	Vehicle Counting	License Plate Recognition	Vehicle Type Classification	Speed Measurement	Parking Management	Weight Measuring
Piezoelectric Sensor	In-Vehicle monitoring the seat occupancy approach. [62]	Indirectly with the combination of vehicle counting and machine learning approach as well as the sensor installing as a part of the vehicles detect crashes. [82,83,84]	This approach uses simple statistical models for analysis electrical signal generated from the mechanical pressure. [85]	Not found publication articles on using this sensor as a solution to this problem at this moment.	It uses an obtained weight features to categorize the vehicles. [85]	Indirectly, installing this system as a vehicle counter in many places in the same road section. [86]	Parking management system solved via vehicle counting entry in park houses. [87]	It uses the feature of increasing weight impacting on strength of generated electrical signals. [88]
FBG sensor	In-Vehicle monitoring the seat occupancy approach. [89,90,91]	Indirectly with the combination of vehicle counting and machine learning approach. [82,83,92]	This approach uses simple statistical models for analysis electrical signal generated from the mechanical pressure. [93]	Not found publication articles on using this sensor as a solution to this problem at this moment.	It uses an obtained weight features to categorize the vehicles. [94]	Signal wave analysis [95].	Parking management system solved via vehicle counting entry in park houses. [96]	It uses the feature of increasing weight impacting on strength of generated electrical signals. [97]
Pneumatic Road Tube	Indirectly via categorization of vehicles proposed for public transport. [98]	Indirectly with the combination of vehicle counting and machine learning approach. [82,83]	This approach uses simple statistical models for analysis electrical signal generated from the mechanical pressure. [98]	Not found publication articles on using this sensor as a solution to this problem at this moment.	It uses an obtained weight features to categorize the vehicles. [98]	Indirectly, installing this system as a vehicle counter in many places in the same road section. [86]	Parking management system solved via vehicle counting entry in park houses. [98]	It uses the feature of increasing weight impacting on strength of generated electrical signals. [99]
Vibration sensor	In-Vehicle monitoring the seat occupancy approach. [100]	Indirectly with the combination of vehicle counting and machine learning approach. [82,83,101]	This approach uses simple statistical models for analysis electrical signal generated from the vibration. [102]	Not found publication articles on using this sensor as a solution to this problem at this moment.	It uses an obtained weight features to categorize the vehicles. [103]	For railway transportation or heavy vehicles, signal processing and analysis are used. [104]	Parking management system solved via vehicle counting entry in park houses. [105]	It uses the feature of increasing weight impacting on strength of generated electrical signals. [106]
Loop Detector	Indirectly via categorization of vehicles proposed for public transport. [107]	Indirectly with the combination of vehicle counting and machine learning approach. [82,83]	This approach uses simple statistical models for analysis electrical signal generated from the magnetic field change. [108]	Not found publication articles on using this sensor as a solution to this problem at this moment.	It uses an obtained weight features to categorize the vehicles. [107]	Indirectly, installing this system as a vehicle counter in many places in the same road section. [86]	Parking management system solved via vehicle counting entry in park houses. [109]	Not found publication article on using this sensor as a solution to this problem at this moment.
Magnetic Sensor	Indirectly via categorization of vehicles proposed for public transport. [110]	Indirectly with the combination of vehicle counting and machine learning approach. [82,83]	This approach uses simple statistical models for analysis electrical signal generated from the magnetic field change. [111]	Not found publication articles on using this sensor as a solution to this problem at this moment.	It uses an obtained weight features to categorize the vehicles. [110]	Indirectly, installing this system as a vehicle counter in many places in the same road section. [86]	Parking management system solved via vehicle counting entry in park houses. [112]	Not found publication articles on using this sensor as a solution to this problem at this moment.
Radio frequency identification	Identification of every passenger approach. [113,114]	Indirectly with the combination of vehicle counting and machine learning approach. [82,83]	This approach reads the data from device [115]	Not found publication articles on using this sensor as a solution to this problem at this moment.	All vehicles identified via this system have to carry information about the type of vehicle implicitly. [116]	Indirectly, installing this system as a vehicle counter in many places in the same road section. [86]	Parking management system solved via vehicle counting entry in park houses. [117]	Not found publication articles on using this sensor as a solution to this problem at this moment.
Acoustic Sensor	Vehicle’s engine sound analysis correlated to vehicle occupancy. [118]	The typical sound of a vehicle collision detects the incident. [119]	This approach uses simple statistical models for analysis electrical signal generated from the sound. [120]	Not found publication articles on using this sensor as a solution to this problem at this moment.	It uses the engine’s sound to categorize the vehicles. [121]	Signal processing and analysis through neural network. [122]	Parking management system solved via vehicle counting entry in park houses. [123]	It uses the feature of increasing weight impacting on strength of generated electrical signals. [124]
Laser Scanner	In-Vehicle monitoring the seat occupancy approach. [125,126]	Indirectly with the combination of vehicle counting and machine learning approach. [82,83]	This approach scans the environment to find vehicles. [127]	Not found publication articles on using this sensor as a solution to this problem at this moment.	It uses spatial parameters information. [128]	It uses the distance change in the time interval to compute the average speed. [129]	The possibility of exact occupancy-free parking slots checking. [130]	Not found publication articles on using this sensor as a solution to this problem at this moment.
Radar	In-Vehicle monitoring the seat occupancy approach. [131]	Indirectly with the combination of vehicle counting and machine learning approach. [82,83]	This approach scans the environment to find vehicles. [132]	Not found publication articles on using this sensor as a solution to this problem at this moment.	It uses spatial parameters information. [133]	It uses the distance change in the time interval to compute the average speed. [79]	The possibility of exact occupancy-free parking slots checking. [134]	Not found publication articles on using this sensor as a solution to this problem at this moment.
Wi-Fi	In-Vehicle monitoring the seat occupancy approach. [135]	Indirectly with the combination of vehicle counting and machine learning approach. [82,83]	This approach uses the signal interuption for detect a vehicle presence. [47]	Not found publication articles on using this sensor as a solution to this problem at this moment.	It uses spatial parameters information. [47]	Indirectly, installing this system as a vehicle counter in many places in the same road section. [86]	The possibility of exact occupancy-free parking slots checking. [136]	Not found publication articles on using this sensor as a solution to this problem at this moment.
Lidar	In-Vehicle monitoring the seat occupancy approach. [137]	The installed sensor could detect the distance that leads to a collision with other object in the traffic. [138]	This approach scans the environment to find vehicles. [139]	Not found publication articles on using this sensor as a solution to this problem at this moment.	It uses spatial parameters information. [140]	It uses the distance change in the time interval to compute the average speed. [141]	The possibility of exact occupancy-free parking slots checking. [142]	Not found publication articles on using this sensor as a solution to this problem at this moment.
Passive Infrared Sensor	In-Vehicle monitoring the seat occupancy approach. [143]	Passive Infrared Sensor could detect a fire. [144]	This approach scans the environment to find vehicles. [145]	Not found publication articles on using this sensor as a solution to this problem at this moment.	It uses spatial parameters information. [146]	Indirectly, installing this system as a vehicle counter in many places in the same road section. [86]	Parking management system solved via vehicle counting entry in park houses. [147]	Not found publication articles on using this sensor as a solution to this problem at this moment.
Ultrasonic Sensor	In-Vehicle monitoring the seat occupancy approach. [148]	The installed sensor could detect the distance that leads to a collision with other objects in the traffic. [101]	This approach scans the environment to find vehicles. [50]	Not found publication articles on using this sensor as a solution to this problem at this moment.	It uses spatial parameters information. [149]	Indirectly, installing this system as a vehicle counter in many places in the same road section. [86]	The possibility of exact occupancy-free parking slots checking. [51]	Not found publication articles on using this sensor as a solution to this problem at this moment.
Camera Systems	Persons counting via computer vision approach on dedicated stations. [61]	Object detection, Object tracking, and image classification approach [150,151].	Object detection and image classification approach [152].	Object detection and image classification aproaches. [153]	It uses spatial parameters information. Image classification and Object Detection [154,155].	Object detection and Object tracking [156].	Object detection and other CV methods [80].	Not found publication articles on using this sensor as a solution to this problem at this moment.

HOV Lane Management

Incident Detection

Vehicle Counting

License Plate Recognition

Vehicle Type Classification

Speed Measurement

Parking Management

Weight Measuring

Piezoelectric Sensor

In-Vehicle monitoring

the seat occupancy approach.

[62]

Indirectly with the combination

of vehicle counting

and machine learning approach

as well as the sensor installing

as a part of the vehicles detect

crashes.

[82,83,84]

This approach uses simple statistical models

for analysis electrical signal generated

from the mechanical pressure. [85]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

It uses an obtained weight

features to categorize

the vehicles. [85]

Indirectly, installing this system

as a vehicle counter in many places

in the same road section. [86]

Parking management system

solved via vehicle counting

entry in park houses. [87]

It uses the feature of increasing

weight impacting on strength

of generated electrical signals. [88]

FBG sensor

In-Vehicle monitoring

the seat occupancy approach.

[89,90,91]

Indirectly with the combination

of vehicle counting

and machine learning approach.

[82,83,92]

This approach uses simple statistical models

for analysis electrical signal generated

from the mechanical pressure. [93]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

It uses an obtained weight

features to categorize

the vehicles. [94]

Signal wave analysis [95].

Parking management system

solved via vehicle counting

entry in park houses. [96]

It uses the feature of increasing

weight impacting on strength

of generated electrical signals. [97]

Pneumatic Road

Tube

Indirectly via categorization

of vehicles proposed

for public transport.

[98]

Indirectly with the combination

of vehicle counting

and machine learning approach.

[82,83]

This approach uses simple statistical models

for analysis electrical signal generated

from the mechanical pressure. [98]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

It uses an obtained weight

features to categorize

the vehicles. [98]

Indirectly, installing this system

as a vehicle counter in many places

in the same road section. [86]

Parking management system

solved via vehicle counting

entry in park houses. [98]

It uses the feature of increasing

weight impacting on strength

of generated electrical signals. [99]

Vibration sensor

In-Vehicle monitoring

the seat occupancy approach.

[100]

Indirectly with the combination

of vehicle counting

and machine learning approach.

[82,83,101]

This approach uses simple statistical models

for analysis electrical signal generated

from the vibration. [102]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

It uses an obtained weight

features to categorize

the vehicles. [103]

For railway transportation

or heavy vehicles,

signal processing and analysis are used. [104]

Parking management system

solved via vehicle counting

entry in park houses. [105]

It uses the feature of increasing

weight impacting on strength

of generated electrical signals. [106]

Loop Detector

Indirectly via categorization

of vehicles proposed

for public transport.

[107]

Indirectly with the combination

of vehicle counting

and machine learning approach.

[82,83]

This approach uses simple statistical models

for analysis electrical signal generated

from the magnetic field change. [108]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

It uses an obtained weight

features to categorize

the vehicles. [107]

Indirectly, installing this system

as a vehicle counter in many places

in the same road section. [86]

Parking management system

solved via vehicle counting

entry in park houses. [109]

Not found publication article

on using this sensor as a solution

to this problem at this moment.

Magnetic Sensor

Indirectly via categorization

of vehicles proposed

for public transport.

[110]

Indirectly with the combination

of vehicle counting

and machine learning approach.

[82,83]

This approach uses simple statistical models

for analysis electrical signal generated

from the magnetic field change. [111]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

It uses an obtained weight

features to categorize

the vehicles. [110]

Indirectly, installing this system

as a vehicle counter in many places

in the same road section. [86]

Parking management system

solved via vehicle counting

entry in park houses. [112]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

Radio frequency

identification

Identification of every

passenger approach.

[113,114]

Indirectly with the combination

of vehicle counting

and machine learning approach.

[82,83]

This approach reads the data from device

[115]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

All vehicles identified

via this system

have to carry information

about the type

of vehicle implicitly. [116]

Indirectly, installing this system

as a vehicle counter in many places

in the same road section. [86]

Parking management system

solved via vehicle counting

entry in park houses. [117]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

Acoustic Sensor

Vehicle’s engine sound analysis

correlated to vehicle occupancy.

[118]

The typical sound of a vehicle

collision detects the incident.

[119]

This approach uses simple statistical models

for analysis electrical signal generated

from the sound. [120]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

It uses the engine’s sound

to categorize the vehicles. [121]

Signal processing and analysis

through neural network. [122]

Parking management system

solved via vehicle counting

entry in park houses. [123]

It uses the feature of increasing

weight impacting on strength

of generated electrical signals. [124]

Laser Scanner

In-Vehicle monitoring

the seat occupancy approach.

[125,126]

Indirectly with the combination

of vehicle counting

and machine learning approach.

[82,83]

This approach scans the environment

to find vehicles. [127]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

It uses

spatial parameters information. [128]

It uses the distance change

in the time interval to compute

the average speed. [129]

The possibility of exact

occupancy-free parking slots

checking. [130]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

Radar

In-Vehicle monitoring

the seat occupancy approach.

[131]

Indirectly with the combination

of vehicle counting

and machine learning approach.

[82,83]

This approach scans the environment

to find vehicles. [132]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

It uses

spatial parameters information. [133]

It uses the distance change

in the time interval to compute

the average speed. [79]

The possibility of exact

occupancy-free parking slots

checking. [134]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

Wi-Fi

In-Vehicle monitoring

the seat occupancy approach.

[135]

Indirectly with the combination

of vehicle counting

and machine learning approach.

[82,83]

This approach

uses the signal interuption

for detect a vehicle presence. [47]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

It uses

spatial parameters information. [47]

Indirectly, installing this system

as a vehicle counter in many places

in the same road section. [86]

The possibility of exact

occupancy-free parking slots

checking. [136]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

Lidar

In-Vehicle monitoring

the seat occupancy approach.

[137]

The installed sensor

could detect the distance

that leads to a collision

with other object in the traffic.

[138]

This approach scans the environment

to find vehicles. [139]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

It uses

spatial parameters information. [140]

It uses the distance change

in the time interval to compute

the average speed. [141]

The possibility of exact

occupancy-free parking slots

checking. [142]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

Passive

Infrared Sensor

In-Vehicle monitoring

the seat occupancy approach.

[143]

Passive Infrared Sensor

could detect a fire.

[144]

This approach scans the environment

to find vehicles. [145]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

It uses

spatial parameters information. [146]

Indirectly, installing this system

as a vehicle counter in many places

in the same road section. [86]

Parking management system

solved via vehicle counting

entry in park houses. [147]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

Ultrasonic Sensor

In-Vehicle monitoring

the seat occupancy approach.

[148]

The installed sensor

could detect the distance

that leads to a collision

with other objects in the traffic.

[101]

This approach scans the environment

to find vehicles. [50]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

It uses

spatial parameters information. [149]

Indirectly, installing this system

as a vehicle counter in many places

in the same road section. [86]

The possibility of exact

occupancy-free parking slots

checking. [51]

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

Camera Systems

Persons counting

via computer vision approach

on dedicated stations.

[61]

Object detection,

Object tracking,

and image classification

approach [150,151].

Object detection and

image classification approach [152].

Object detection

and image classification aproaches.

[153]

It uses

spatial parameters information.

Image classification

and Object Detection [154,155].

Object detection

and Object tracking [156].

Object detection

and other CV

methods [80].

Not found publication articles

on using this sensor as a solution

to this problem at this moment.

Table 2 describes the potential of mentioned sensors to solve transportation monitoring tasks based on motion detection and object tracking. From the sensors mentioned, it can be concluded that we may solve many monitoring transportation tasks via the camera system in almost all applications except for weight measurement. CV methods and algorithms enable the creation of a virtual sensor (Figure 1) from the camera to sense many objects. The research’s progress in traditional CV methods and artificial deep neural networks makes motion-detection and object-tracking methods more effective [157,158].

Intrusive sensors such as magnetic sensors, loop detectors or pneumatic road tubes use a specific installation configuration for each purpose (Table 2). This installation configuration impacts the detector sensitivity level. Each monitoring task requires a different sensitivity level. For example, single- and dual-loop detector configurations have different sensitivity levels. The parking slot management system does not need to categorize the vehicle; therefore, the single-loop detector is efficient for parking slot management. In monitoring highways, the information about counted trucks and other vehicle types has a significant role. Dual-loop detectors increase the vehicle categorization accuracy [41].

FBG sensors are the most prospective intrusive sensors (Table 2). Suitable installation configurations and machine learning models can classify many events. The disadvantage is a calibration system. The advantage of the FBG sensor is the low-cost manufacturing investments compared with metallic sensors, long service life, and low maintenance cost. The FBG sensor is categorised as a virtual sensor source for traffic monitoring tasks in intelligent transportation systems because it measures a reflected light spectrum due to the fibre strain changes. It requires a complex, trained machine learning model to filter and extract valuable data for vehicle presence, speed estimation or weight measurement [32,159].

This article proposes categorising intrusive sensors, except FBG sensors, into physical sensors because the installation configuration classifies the vehicle presence directly via simple signal processing. These sensors consider common vehicle features that every vehicle consists of metal parts. The metal parts change the magnetic field’s behaviour (magnetometer, loop detector) or impact the vehicle’s weight characteristic (pneumatic road tube) [41].

In conclusion, intrusive sensors except FBG sensors have limited usage for collecting complex data about the actual situation on the road.

Referring to Table 2, intrusive sensors such as loop detectors or magnetometers are limited to use in the HOV lanes management system. However, almost all vision-based sensors can check a vehicle’s occupancy status. Despite the fact that vision-based sensors are suitable for this purpose, there are still some problems. Lidars and laser scanners are too expensive for installing to the car’s interior only for counting the passengers. An automatic passenger-counting system based on a camera system needs to be installed at distances from which the person’s face could be recognizable. This approach is intrusive. The optimal solution is using sensors detecting passengers via seat occupation and sending the passenger count from the vehicle with V2I protocol to the HOV lane station.

The parking assistance system is one of the standard systems being implemented in cars. This system helps the driver search for a free parking slot and successfully park without collision. Parking houses can monitor free parking slots with a huge number of intrusive sensors that are costly to implement and maintain or with few cameras which can cover the same area with cheaper investment. Cameras can navigate cars across a parking lot and help calculate the trajectory necessary to reach a free parking slot. Parking management systems will become a vital part of the intelligent transportation system, with which V2I will cooperate [80,165,166,167].

Intrusive sensors have limitations in gaining complex data about the actual situation in monitoring road tasks. All intrusive sensors have a smaller distance detection range than non-intrusive sensors. These sensors can reliably categorize and count vehicles, and some of them, such as a magnetometer or FBG sensor, can measure speed. On the other hand, the expensive installation process limits their multiple usages in the same route. The most significant advantage of intrusive sensors is that they can only measure a vehicle’s weight. However, not all intrusive sensors can measure a vehicle’s weight, for example, loop detectors or magnetometers.

FBG sensors can monitor almost all tasks mentioned in (Table 2) except the ALPR task. This sensor can be used as a universal intrusive virtual hardware-based sensor for intelligent transportation systems tasks because it can solve most tasks from intrusive sensors (Table 2).

Camera systems visually observe roads by making video records. CV methods can extract a lot more information from these video records. High-resolution images or videos increase details and data quality for the following analysis process. Optimized convolutional neural network (CNN) models detect many objects of interest via classification rectangle (object detection methods) [168]. Object-detection methods are used for almost all monitoring tasks (Table 2). The CV methods that run on the camera system videos form a universal non-intrusive virtual sensor in the intelligent transportation system (Table 2).

This entry is adapted from the peer-reviewed paper 10.3390/s23094469

© Text is available under the terms and conditions of the Creative Commons Attribution (CC BY) license; additional terms may apply. By using this site, you agree to the Terms and Conditions and Privacy Policy.