Enhancing Pavement Condition Assessment: A Comprehensive Review

of Affordable Sensing Technologies for Cycle Tracks

Jeziel Antonio Ayala Garcia

1 a

, Syed Shah

1 b

, Muhammad Hassam Baig

1 c

Waqar Shahid Qureshi

1 d

and Ihsan Ullah

2 e

School of Computer Science, University of Galway, Galway, Ireland

Insight SFI Research Centre for Data Analytics, University of Galway, Galway, Ireland

{j.ayalaGarcia1, waqarshahid.qureshi, ihsan.ullah, syed.shah, muhammadhassam.baig}@universityofgalway.ie

Keywords:

Low-Cost Data Collection, Sensors, Mobile Phones, Pavement Condition Assessment, Cycling Tracks.

Abstract:

This paper explores the potential of low-cost sensing technologies for assessing the condition of cycling track

pavement. As cycling gains popularity, the demand for efﬁcient pavement maintenance solutions increases.

Machine survey-based pavement condition assessment often relies on expensive, specialized equipment, which

is not always suitable for cycle tracks due to limited budgets and accessibility, hence the need for low-cost so-

lutions. The integration of low-cost sensing technologies and data collection, such as inertial measurement

units (IMUs), low-cost imaging sensors, and crowdsourced data collection, presents a promising alternative

for cycle track pavement surface assessment. This paper highlights the advantages of employing out-of-the-

shelf devices such as smartphones with GPS, cameras, and IMU, or action cameras with GPS and IMU to

collect images, their location, and vehicle vibration in a speciﬁc orientation. This data is then used to estimate

pavement roughness, detect surface distress, and compute pavement surface condition. Literature reviews

reveal a signiﬁcant gap in the utilization of these technologies for cycle tracks, suggesting a promising area

for further research and application. Furthermore, it proposes a software framework for data collection and

visualization to combine these technologies to enhance the efﬁciency and reliability of pavement assessment

for cycling infrastructure at a lower cost than machine-based pavement assessment and faster and more quan-

titative than manual surveys. This paper emphasizes the need for more practical and scalable solutions that

support the maintenance of sustainable transportation systems.

1 INTRODUCTION

Pavement condition assessment can allow the right al-

location of resources in the decision-making for main-

tenance according to FHWA of the US (Federal High-

way Administration (FHWA), nd), this applies to mo-

tor roads but also the same could be said about cycling

tracks.

Current automated pavement condition assess-

ment requires specialized equipment, which is mainly

used for pavement motor roads. This could be costly

and the vehicles called proﬁlers might not be suitable

for cycling tracks. Technology advances have made it

possible to use low-cost technologies as an alternative

https://orcid.org/0009-0006-7829-9948

https://orcid.org/0009-0008-5653-2045

https://orcid.org/0009-0003-3153-9527

https://orcid.org/0000-0003-0176-8145

https://orcid.org/0000-0002-7964-5199

to specialized equipment for automated pavement as-

sessment, such as a mobile phone with IMU sensors,

to assess the quality of pavement by obtaining the

Roughness Index (RI) (Jeong et al., 2020) or to use

machine learning to assess the quality of pavement

using images such as (Shin et al., 2024). Still, most

of this work has been done for motor tracks, leaving

a great opportunity to explore these alternatives for

pavement assessment applied to cycling tracks.

Looking forward to sustainable alternatives in

transportation, bicycle infrastructure gains more rel-

evance. However, tracks in bad condition represent a

risk for current users and could undermine the inter-

est of other users in choosing these methods of trans-

portation that require using bicycle tracks in good

condition. Often, the budget for maintenance is lim-

ited, so a low-cost pavement assessment for cycling

tracks could be the solution (Massow et al., 2024).

As a low-cost alternative in this paper, it is pro-

posed the integration of different devices as sensing

676

Garcia, J. A. A., Shah, S., Baig, M. H., Qureshi, W. S. and Ullah, I.

Enhancing Pavement Condition Assessment: A Comprehensive Review of Affordable Sensing Technologies for Cycle Tracks.

DOI: 10.5220/0013505100003941

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 11th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2025), pages 676-682

ISBN: 978-989-758-745-0; ISSN: 2184-495X

devices such as mobiles and action cameras, devices

that have integrated GPS, camera, and IMU sensors,

with the use of a mobile app for data collection, a

web app for visualization and server for data process-

ing and storing, all of this to make a software frame-

work for pavement assessment specialized on cycling

tracks. This system was tested and thought to be used

in the Republic of Ireland, therefore the tests were

conducted on ofﬂine cycling tracks, dedicated cycling

paths that are physically separated from motor roads,

denominated like this by Transport Infrastructure Ire-

land, (Transport Infrastructure Ireland, 2022).

2 REVIEW OF EXISTING

TECHNOLOGIES

2.1 Overview of Pavement Condition

Assessment

Users determine a road’s pavement condition by its

ride quality. Smoothness speciﬁcations are com-

monly used to assess this, for example, indicating the

roughness level on the pavement surface. One param-

eter widely used to judge the condition of the pave-

ment is to measure the smoothness of it, for this, a

metric widely used is the International Roughness In-

dex (IRI). To obtain this measure usually, specialized

equipment such as an inertial proﬁler is used, which

is a vehicle equipped with an inertial proﬁling sys-

tem that includes accelerometers, height sensors, a

distance design system, and a computer to obtain the

longitudinal proﬁle of the road from where the IRI is

calculated (Board et al., 2018). If we would like to

replicate the process for obtaining the IRI we would

require a specialized vehicle to assess ofﬂine cycling

tracks since the road width is not enough to ﬁt most

of the proﬁler-vehicles used for motor road tracks.

Identifying distresses that occur on the pavement

is relevant for assessing pavement condition, such as

rutting, and cracking, manual surveys can be done to

identify them along the segment of road selected for

the inspection, although this process is limited to just

segments of the road, and not the entire of the road

infrastructure. Rutting is a signiﬁcant failure that can

occur due to mechanical loading and thermal varia-

tion due to climatic effects that, over time, cause de-

formation (rutting) on the material (Perraton et al.,

2011). Another distress relevant for pavement assess-

ment is cracking, whether caused by fatigue or ther-

mal cause, a relevant issue that has to be addressed by

a proper road maintenance system (Abed et al., 2023).

Another parameter relevant for pavement assess-

ment is pavement texture due to its correlation to

asphalt pavement adhesion. Pavement texture plays

an important role by providing skid resistance, ul-

timately providing driving safety by allowing users

to have the grip necessary in case of an emergency

brake. To measure pavement adhesion, special types

of equipment have been developed, whether to mea-

sure the braking force directly using devices that need

testing tires or rubber slides and water for doing

the test or by calculating the depth of the road tex-

ture by different methods such using a tracking me-

ter or a highspeed proﬁler to obtain the Mean Proﬁle

Depth, and most of these devices require manual op-

eration making them time-consuming and not feasible

to large scale due to being labor intensive (Liang et al.,

2024). Therefore, the necessity for cost-effective al-

ternatives emerges, which will be examined in the fol-

lowing discussion.

2.2 Low-Cost Sensing Technologies

Since budget limitation is a common problem in asses

pavement for cycling tracks (Massow et al., 2024), we

propose to address it with a software framework that

involves an integration of low-cost sensing technolo-

gies and an efﬁcient data collection technique such a

crowdsourcing while addressing the limitations that a

single low-cost sensing technology might have by it-

self alone, technologies such as using inertial sensors,

image-based techniques, by using a combination of

them in a system that integrates the selected sensors

to complement the insights of the different data ob-

tained from them such as images, location, vibration

and direction of the vehicle.

2.2.1 Inertial Sensors

Research has been conducted to use Inertial Measure-

ment Unit (IMU) sensor signals to extract data to de-

scribe the roughness of the road, demonstrating that

the use of GPS/IS systems provides valuable informa-

tion on the roughness of the road. However, it is worth

mentioning that obtaining the most precise readings

might not be possible since the objective is to use out-

of-the-shelf devices. Still, by changing from the time

domain, which is more susceptible to noise, using the

Fourier transform, it is possible to change to the fre-

quency domain, which allows the extraction and re-

motion of noise (Wen, 2008), still after this quality of

the data obtained is not as precise as other automated

pavement assessment methods such a using a proﬁler.

It is worth mentioning that this work was done and

tested on automobiles on motor roads and not on cy-

cling tracks with the different types of vehicles that

can use this type of infrastructure.

Enhancing Pavement Condition Assessment: A Comprehensive Review of Affordable Sensing Technologies for Cycle Tracks

677

More recent studies have obtained great results

comparing readings from an IMU Raspberry-Based

device with a road surface proﬁler, proving the fea-

sibility of using these types of inertial sensors to

do considerable low-cost surveys on pavement as-

sessment, taking into consideration that most smart-

phones at the market have these sensors available, al-

though is worth mentioning that even that progress

is made in this topic the intention is not to put this

kind of devices at the same precision level of dedi-

cated devices, instead the goal is providing a practical

and low-cost alternative (Loprencipe et al., 2021).

The use of inertial sensors by itself might not

be enough; therefore, applying a Convolutional Net-

work Method using data obtained from IMU sensors

can be used to detect potholes with higher accuracy

(Ozoglu and G

okg

oz, 2023), who proposed a novel

CNN model that can be used in the context of Road

Surface Recognition.

As mentioned before, IRI is one the most impor-

tant pavement assessment metrics therefore, the im-

portance obtaining this measure by utilizing the sen-

sors within smartphones can be used to obtain IRI, al-

though there could be some uncertainty on the results

due to slope variance, different types of pavement dis-

tress, the type to the testing vehicle it is possible to re-

duce the errors through signal processing techniques

such as moving average ﬁlter, Butterworth bandpass

ﬁlter, and baseline correction ﬁlter to the accelera-

tion data obtained (Alatoom and Obaidat, 2022). It is

mentioned that the type of mounting for the mobile,

the type of vehicle, and the speed affects the readings,

which is a matter of importance since the type of ve-

hicles that can use a cycling track have important dif-

ferences from automobiles.

Also, it is important to mention that the tires do

not cover the full width of the pavement and much rel-

evant information is lost or not considered while rec-

ollecting data, especially on cycling tracks since most

of the vehicles that use this type of infrastructure such

as cycles and scooters, like the vehicles used for our

tests had only two wheels covering much less surface

that a four-wheel will do, therefore to address this will

discuss about image-based techniques.

2.2.2 Image-Based Techniques

The Pavement Condition Assessment (PCI) is used as

a metric where distresses are identiﬁed. This method

mainly relies on manual methods to obtain this met-

ric, which could be costly and prone to human errors.

This could be addressed by the method proposed by

(Ibragimov et al., 2024) a method that relies on deep

learning and image processing technologies to detect

and analyse cracks for the later PCI calculation. Sim-

ilarly (Nasrallah et al., 2024) used image processing

to identify cracks but also integrated a Global Naviga-

tion System along images recorded by phones, an ap-

proach proposed that automates the process of crack

inspection, allowing identiﬁcation and locating dis-

tress.

As mentioned before image processing with the

aid of machine learning can be used for crack detec-

tion on pavement surfaces, from common algorithms

such as linear regression or more advanced ones such

as using mask R-CNN predictive models, allowing its

use for automation in pavement management systems,

where image classiﬁcation for categorizing reﬂective

cracking in clear weather conditions has achieved an

accuracy of 95.9% by the approach proposed by (Shin

et al., 2024).

Deep learning can be applied to pavement as-

sessment, like the YOLO9tr (Youwai et al., 2024),

a model for pavement assessment that is capable of

correctly identifying pavement distresses from video

up to 136 FPS, making it feasible the use of this sys-

tem for automated inspections, allowing the integra-

tion this model in a system for pavement maintenance

(Youwai et al., 2024). Although these advances show

the potential of low-cost image sensors to automate

pavement assessment, these methods still have some

limitations such as conﬁguration from where the cam-

era needs to take the images of the track in an orthog-

onal position for the correct detection of the distresses

on the pavement surface while collecting data, as well

that these projects have been tested only on automo-

biles and not on bicycles, or any other vehicles that

can use cycling tracks.

2.2.3 Crowdsourced Data

Data obtained from smartphones for pavement assess-

ment has been addressed, and previous research stud-

ies have obtained ways for calculating IRI without

a restrictive controlled environment setting, proving

feasible the gathering helpful information for road in-

frastructure; what is more, mobile crowdsourcing data

could be an effective way of collecting information

for a Pavement Management System. Although do-

ing this presents a lot of challenges, such as the varia-

tion vehicles, variance in the sensor of different smart-

phones, different ways of mounting the smartphone,

and many other variables. These variables need to

be addressed to obtain a robust performance of the

implementation, like the CNN for IRI estimation by

(Jeong et al., 2020) that was validated in the crowd-

sourced experiment where other challenges have been

addressed, such as the GPS data variations, but the re-

sults showed on the crowdsourced IRI estimation by

(Jeong and Jo, 2024) presents great potential for sim-

VEHITS 2025 - 11th International Conference on Vehicle Technology and Intelligent Transport Systems

678

ilar applications where smartphones or any other out-

off-the-shelf devices can be used for collecting data

such a Pavement Assessment Management system for

cycling tracks.

Cycling tracks can use the crowdsourced data col-

lection for pavement assessment, acknowledging the

limitations such as the usually low budget and that

pavement assessment for cycling is not as regulated

as it is for pavement for motor roads, the use of sens-

ing technologies such as the IMU sensor within mo-

bile phones, that indeed is less precise than other

automated processes that require specialized equip-

ment used for pavement assessment in motor roads,

it makes the use low-cost technologies more rele-

vant. The development of a dedicated application

for the collection of data can allow the implementa-

tion of crowdsourced systems for pavement assess-

ment (Massow et al., 2024), although in the crowd-

sourced work done by Massow they concentrate on

the collection of vibration and direction of vehicles

using IMU presenting various alternatives for mea-

suring the roughness of the surface, the fact that the

image-based sensors where not implemented leave an

opportunity of enhancing this methodology.

3 THE PROPOSED SOFTWARE

FRAMEWORK

Previous approaches have taken crowdsourcing as a

way to make an efﬁcient, low-cost pavement assess-

ment system for bicycle tracks (Massow et al., 2024);

still, they discarded using image techniques due to

the technical difﬁculties that implementing this could

have; we believe that with current technology and

knowledge in the ﬁelds image processing and com-

puter vision can allow the use of videos for assessing

pavement condition in cycling tracks, although in this

case, we will not concentrate on this process, leaving

it for future work, will concentrate on the method-

ology for collecting data. In previous research, the

use IMU sensing devices such as smartphones can be

used to obtain valuable information for determining

the condition of the pavement using metrics such as

the IRI (Jeong et al., 2020), (Massow et al., 2024).

Designing a low-cost Pavement Management System

based on the information obtained from IMU sens-

ing devices and Image sensing devices together can

bring a better understanding of the condition of a spe-

ciﬁc bicycle route infrastructure monitored through

the system. The system we propose consists of a mo-

bile application app that works as the input of data, the

information is collected from an action camera or mo-

bile phone mounted on a bicycle, a server from which

the data captured from our mobile app will be pro-

cessed to be interpretable for pavement assessment,

and a web application that would serve as a monitor-

ing tool where a map with relevant information about

the cycling tracks monitored will be presented, this

displayed on Figure 1.

We deﬁned our two main types of data-collecting

devices: action cameras and mobile devices. Both

types of devices have inertial sensors such as ac-

celerometers, GPS, gyroscopes, or magnetometers, as

well as a camera that can record video and sensor

readings simultaneously. For example, it is possible

to create a device that uses both kinds of information

to describe the condition of the cycling track’s pave-

ment stretch.

3.1 Data Extraction

3.1.1 Action Camera Data Extraction

For this project, the GoPro Black 11 and GoPro

Model 9 cameras were selected. These models embed

sensor readings within MP4 ﬁles, allowing video, ac-

celerometer, gyroscope, and GPS data to be recorded

simultaneously if set up correctly. To capture data,

the camera or phone must be mounted at the han-

dlebar center, pointing at the road. The open-source

GPMF-parser library (GoPro, 2024) was modiﬁed to

extract accelerometer, gyroscope, and GPS readings

from MP4 ﬁles into CSVs for processing. Initial tests

used only GoPro 11 Black videos of Irish cycling

tracks, as it features GPS9, a newer and more precise

sensor. Later, GoPro 9 Black videos were added, re-

quiring modiﬁcations to handle its older GPS5 sensor

alongside GPS9.

3.1.2 Mobile Data Extraction

One problem to deal with is that there is much vari-

ance within the type of sensors that every mobile

phone could or could not have since, for this exper-

iment, we used a Samsung A15 and a Xiaomi Redmi

12C. In contrast, the ﬁrst has an accelerometer, GPS,

and magnetometer; the second device only has an ac-

celerometer and GPS, which is not enough for cal-

culating IRI since we may need data from a gyro-

scope or magnetometer. We can observe that not ev-

ery smartphone on the market will have the necessary

sensors to obtain an approximation of the IRI. How-

ever, we can record the readings from a magnetome-

ter and gyrometer, alongside GPS, accelerometer, and

video, saving data from each sensor into a CSV ﬁle in

the same format used for the one recording data from

action cameras, considering that only a gyrometer or

magnetometer to calculate the IRI is needed, but by

Enhancing Pavement Condition Assessment: A Comprehensive Review of Affordable Sensing Technologies for Cycle Tracks

679

Figure 1: Software Framework for Pavement Assessment for Cycling Tracks.

having the two of them we increase the number of de-

vices that can be used to record pavement condition

through our dedicated app.

Throughout the development period of the appli-

cation, we held meetings with stakeholders, such as

industry-related experts and active users of cycling

tracks, and much feedback has been received. One of

the things we ended up implementing from this was

a manual way of reporting pavement distresses such

as potholes, cracks, debris, ﬂoodings, and or invading

vegetation through images that are linked to the GPS

coordinates at the moment of a picture is taken, allow-

ing to report these distresses on real-time and being

able to display them on the web application cycling

track map.

3.2 Mobile Application

The mobile application was developed in Kotlin using

Android Development Studio; therefore, the current

version of our app is compatible with Android De-

vices. Jetpack Compose was used to develop the user

interface. As this app is to be used in a crowdsourced

pavement assessment management system, we need

to manage user authentication services. To do this,

Firebase Authentication services were used, allowing

us to comply with user data protection in the EU.

As of the publication of this paper, the current

version of the app has four main menus: the home-

page, the recording menu, the action camera connec-

tion menu, and the ﬁle manager menu. The home

menu, apart from the basic functionality of any other

mobile application, includes instructions and settings

Figure 2: Mobile Application camera Recording Menu and

Home Menu. The mobile application records video, GPS,

and IMU sensors data simultaneously.

that could show new users how to start recording and

uploading data to the server.

The recording menu has two main functionalities,

which can be visualized in Figure 2. The ﬁrst one al-

lows users to record video, accelerometer data, mag-

netometer, and gyrometer simultaneously while al-

lowing users to declare from which type of vehicle the

data is going to be recorded; in this menu, the video

and if sensors available on the mobile are streamed in

the screen while recording.

VEHITS 2025 - 11th International Conference on Vehicle Technology and Intelligent Transport Systems

680

The second functionality implemented in this

menu is to manually record visual distress on the

pavement by allowing users to take pictures and se-

lect the type of distress they want to report. The user

interface on this menu has been improved through the

continued feedback from our stakeholders during our

tests and meetings. The action camera connection

serves as a bridge for downloading videos or pictures

taken from an action camera, extending the number of

devices that can used to record data from more users.

This can also be done using the ﬁle manager menu

to upload videos taken from an action camera down-

loaded to mobile devices through different sources.

The ﬁle manager menu deals with the important

task of sending the data to the server; having to send

data manually instead of sending real-time data over-

comes the problems of dealing with signal problems

or battery drowning and having the option to send the

data afterward. This allows users to record data even

when ofﬂine since the application was thought to be

used on cycling routes where there could be a lousy

internet signal.

3.3 Visualization App

The visualization web serves as the monitoring tool

of the cycling pavement infrastructure. A map devel-

oped using MapBox shows the condition of the as-

sessed cycling tracks; the information is displayed,

showing the metric obtained from the IMU sensors

and the pavement rating based on a visual inspection.

This could be done through a rating designed to as-

sess the condition of cycling tracks, whether done by

industry experts or AI.

To achieve the classiﬁcation on the visualization

app, it was use the Cycle Route Surface Index (CRSI)

proposed in (Shah et al., 2025) for classifying the bi-

cycle route, an index that rates the condition of the

route from 1 to 5, ﬁve meaning the best and one the

worst condition for utilizing the cycling route, during

the pilot testing different deep learning models were

used to classify on CRSI, that were trained using the

data labeled by experts from Transport Infrastructure

Ireland, refer to Shah work for more details.

With regard to the IRI displayed on the visualiza-

tion app it was obtained from an early version of the

method developed by (Baig et al., 2025). This method

uses the data obtained from the data collected and ex-

tracted described in the data extraction section, the

data once received on the server is used to calculate

the IRI.

Both IRI and CRSI values are synchronized with

speciﬁc frames by linking them to the closest fre-

quency measurement. This approach accounts for the

varying data rates—accelerometer and gyro at 200

Hz, video at 60 Hz, and GPS at 10 Hz—with all data

ultimately aligned at the 10 Hz interval dictated by the

GPS, showing each frame linked to the GPS location

of the route on the map.

4 FUTURE DIRECTIONS

The implementation of a monitoring system needs the

active collaboration of local authorities for correct im-

plementation since administrative decisions can mod-

ify the way of the designed system.

Future experiments and data collection using spe-

cialized equipment, such as an inertial proﬁler, to

compare the data obtained from our mobile app and

action cameras and ﬁnd the correlation between them;

although similar experiments have been done in the

past, current investigation on a dedicated method de-

veloped for processing the data to obtain IRI, which

needs to be tested.

The initial data from which initial experiments

were done was taken from bicycles provided by TFI,

this is relevant since the cycling tracks are not used

only by this type of vehicle, therefore on the most re-

cent experiments the data was recorded from electric

bicycles, electric scooters, and bicycles, these results

need to be compared against each other since also data

was recorded using different mobiles devices on dif-

ferent vehicles.

5 CONCLUSIONS

The review demonstrates the viability of low-cost

sensing technologies as practical alternatives for as-

sessing cycling track pavements, especially in the

context of budget constraints and the need for scalable

solutions. By integrating GPS, IMU, and image sen-

sors, image processing, and crowdsourced data col-

lection applied to pavement assessment, these meth-

ods offer a robust framework that can give substan-

tial information on the current condition of cycling

tracks. The proposed integration of a mobile appli-

cation, off-the-shelf sensing devices, and a computer-

ized monitoring system enables data collection and

visualization, laying the groundwork for improved

maintenance strategies in the context of sustainable

transport alternatives such as cycling tracks.

While these technologies are not intended to re-

place specialized equipment, their accessibility and

cos-effectiveness make them invaluable for large-

scale applications promoting a safer, well-maintained

cycling infrastructure. Future work should focus on

Enhancing Pavement Condition Assessment: A Comprehensive Review of Affordable Sensing Technologies for Cycle Tracks

681

validating and reﬁning the proposed system through

more ﬁeld experiments, exploring correlations with

measurement devices such as proﬁlers, and obtain-

ing a reliable pavement assessment measure such as

IRI while addressing implementation and continuing

with stakeholder collaboration. Adopting such sys-

tems can play a pivotal role in advancing sustainable

transportation by ensuring cycling tracks remain safe

and appealing to users.

ACKNOWLEDGMENTS

This research was conducted with ﬁnancial support

from the EU Commission’s Recovery and Resilience

Facility through the Research Ireland OurTech

Challenge (Grant Number 22/NCF/OT/11220)

and Science Foundation Ireland (Grant Number

SFI/12/RC/2289 P2) under the Insight SFI Research

Centre for Data Analytics. The authors also acknowl-

edge the support of Transport Infrastructure Ireland

and Katleen Bell-Bonjean, Social Impact Champion

from GORTCYCLETRAILS.ie. For Open Access, a

CC BY public copyright license has been applied to

any Author Accepted Manuscript resulting from this

submission.

REFERENCES

Abed, A., Rahman, M., Thom, N., Hargreaves, D., Li, L.,

and Airey, G. (2023). Predicting pavement perfor-

mance using distress deterioration curves. Road Ma-

terials and Pavement Design.

Alatoom, Y. I. and Obaidat, T. I. (2022). Measurement of

street pavement roughness in urban areas using smart-

phone. International Journal of Pavement Research

and Technology, 15(5):1003–1020.

Baig, M. H., Ayala Garcia, J. A., Qureshi, W. S., and Ullah,

I. (2025). Towards assessing cycleway pavement sur-

face roughness using an action camera with imu and

gps.

Board, T. R., National Academies of Sciences, E., and

Medicine (2018). Inertial Proﬁler Certiﬁcation for

Evaluation of International Roughness Index. The Na-

tional Academies Press, Washington, DC.

Federal Highway Administration (FHWA) (n.d.). Practical

guide for quality management of pavement condition

data collection. U.S. Department of Transportation.

GoPro (2024). Gpmf-parser: Parser for gpmf™ formatted

telemetry data used within gopro® cameras. GitHub

repository. Retrieved from GitHub.

Ibragimov, E., Kim, Y., Lee, J. H., Cho, J., and Lee, J.-J.

(2024). Automated pavement condition index assess-

ment with deep learning and image analysis: An end-

to-end approach. Sensors, 24(7):2333.

Jeong, J.-H. and Jo, H. (2024). Toward real-world

implementation of deep learning for smartphone-

crowdsourced pavement condition assessment. IEEE

Internet of Things Journal, 11(4):6328–6337.

Jeong, J.-H., Jo, H., and Ditzler, G. (2020). Con-

volutional neural networks for pavement roughness

assessment using calibration-free vehicle dynamics.

Computer-Aided Civil and Infrastructure Engineer-

ing, 35(11):1209–1229.

Liang, H., Pagano, R. G., Oddone, S., Cong, L., and

De Blasiis, M. R. (2024). Analysis of road surface

texture for asphalt pavement adhesion assessment us-

ing 3d laser technology. Remote Sensing, 16(11).

Loprencipe, G., de Almeida Filho, F. G. V., de Oliveira,

R. H., and Bruno, S. (2021). Validation of a low-cost

pavement monitoring inertial-based system for urban

road networks. Sensors, 21(9).

Massow, K., Maiwald, F., Thiele, M., Heimendahl, J.,

Protzmann, R., and Radusch, I. (2024). Crowd-based

road surface assessment using smartphones on bicy-

cles. In 2024 International Conference on Artiﬁcial

Intelligence, Computer, Data Sciences and Applica-

tions (ACDSA), pages 1–8, Victoria, Seychelles.

Nasrallah, A. A., Abdelfatah, M. A., Attia, M. I. E., and El-

Fiky, G. S. (2024). Positioning and detection of rigid

pavement cracks using gnss data and image process-

ing. Earth Science Informatics, 17:1799–1807.

Ozoglu, F. and G

okg

oz, T. (2023). Detection of road

potholes by applying convolutional neural network

method based on road vibration data. Sensors,

23(22):9023.

Perraton, D., Di Benedetto, H., Sauz

eat, C., Bankowski, W.,

Partl, M., and Grenfell, J. (2011). Rutting of bitumi-

nous mixtures: wheel tracking tests campaign analy-

sis. Materials and Structures, 44(5):969–986.

Shah, S. M. H., Qureshi, W. S., O’Dea, G., and Ullah, I.

(2025). Intelligent pavement condition rating system

for cycle routes and greenways.

Shin, S.-P., Kim, K., and Le, T. H. M. (2024). Feasi-

bility of advanced reﬂective cracking prediction and

detection for pavement management systems using

machine learning and image detection. Buildings,

14(6):1808.

Transport Infrastructure Ireland (2022). Rural cycleway de-

sign (ofﬂine & greenway) (dn-geo-03047). Transport

Infrastructure Ireland. Retrieved from Transport In-

frastructure Ireland.

Wen, W. (2008). Road roughness detection by analysing

imu data. Master’s thesis, School of Architecture and

the Built Environment, Royal Institute of Technology

(KTH), Stockholm, Sweden.

Youwai, S., Chaiyaphat, A., and Chaipetch, P. (2024).

Yolo9tr: A lightweight model for pavement damage

detection utilizing a generalized efﬁcient layer aggre-

gation network and attention mechanism. Journal of

Real-Time Image Processing, 21(5).

VEHITS 2025 - 11th International Conference on Vehicle Technology and Intelligent Transport Systems

682