Optimizing UAV-Based Inventory Detection and Quantification in

Industrial Warehouses: A LiDAR-Driven Approach

SOTIRIOS TSAKIRIDIS1, APOSTOLOS PAPAKONSTANTINOU2, ALEXANDROS KAPANDELIS1,

PARIS MASTOROCOSTAS3, ALKIVIADIS TSIMPIRIS1, DIMITRIOS VARSAMIS1

1Department of Computer, Informatics and Telecommunications Engineering

International Hellenic University

Serres Campus

GREECE

2Department of Civil Engineering and Geomatics

Cyprus University of Technology

Lemesos

CYPRUS

2Department OF Informatics and Computer Engineering

University of West Attica

Egaleo

GREECE

Abstract: The advancement of technology has brought about a revolution in industrial operations, where special-

ized tools play a crucial role in enhancing efficiency. This study delves into the significant impact of the logistics

department in global industries and proposes an innovative solution for inventory detection and recognition using

unmanned aerial vehicles (UAVs) equipped with LiDAR technology. Unlike existing research that often involves

intricate hardware systems and algorithms leading to increased costs and computational demands, our research

focuses on streamlining the inventory detection process by utilizing a LiDAR data and an algorithmic approach

that minimizes the time of extensive counting process into the warehouse to quantify the pallets existing. The pro-

posed methodology entails a custom-made quadcopter equipped with a single-beam and high-frequency LiDAR

range finder. Operating autonomously along a predetermined flight plan, the drone captures high-frequency range

data of warehouse inventory. The paper comprehensively outlines the UAV control procedures, warehouse scan-

ning using LiDAR, and the inventory detection and quantification of pallets algorithmic process. The proposed

method processes LiDAR data in a post-process way, estimating the number of pallets and, consequently, produc-

ing a map of each stack within the warehouse denoting the quantities of pallets. The research results showcase

the successful implementation of the proposed approach in a model warehouse, achieving an impressive 100%

evaluation accuracy. Future research endeavors aim to extend this methodology to warehouses with dynamic

product placements, emphasizing real-time monitoring for comprehensive inventory detection. This innovative

approach stands out as a cost-effective and efficient solution for industries seeking accurate and timely inventory

information.

Key-Words: LiDAR, UAVs, Inventory Detection, Warehouse Measurement

Received: June 23, 2023. Revised: December 9, 2023. Accepted: January 11, 2024. Published: February 27, 2024.

1 Introduction

The evolution of technology in our days has brought

significant progress and improvement in every sec-

tor of the industry area. Specialized and constantly

evolving technological tools are being incorporated

daily into industrial units to facilitate the implemen-

tation of industrial production, organize a set of tasks

necessary to be carried out within the operations

framework of an industrial unit, optimize the devel-

opment and marketing departments of the industries,

and achieve the most immediate and effective ”digi-

tal” connection between different departments of an

industry unit or among units within a global industry

(communication between branches on a global scale).

As a result, the continuous integration of new tech-

nologies into industrial units is critical and obsolete

this is administered through new high-end hardware

implementation or new and improved software inte-

gration in each department of any structured industry.

Unmanned aerial vehicles (UAVs), commonly re-

ferred to as drones, have garnered considerable atten-

WSEAS TRANSACTIONS on SYSTEMS

DOI: 10.37394/23202.2024.23.14

Sotirios Tsakiridis, Apostolos Papakonstantinou,

Alexandros Kapandelis, Paris Mastorocostas,

Alkiviadis Tsimpiris, Dimitrios Varsamis

E-ISSN: 2224-2678

121

Volume 23, 2024

tion and found widespread commercial applications

across various domains, [1], [2]. UAVs, with their dis-

tinctive characteristics, offer specific advantages tai-

lored to focused tasks or activities. In recent years,

several studies have introduced innovative initiatives

to foster the adoption and implementation of drones

in warehouse management. Notably, these initiatives

encompass the integration of Unmanned Aerial Ve-

hicles (UAVs) with technologies such as barcodes,

computer vision (CV), and machine learning (ML),

[3]. These integrations are designed to automate var-

ious warehouse operations, including tasks like bar-

code detection and decoding, [4], [5]. Simultane-

ously, studies, such as [6], have explored enhanc-

ing self-positioning UAVs and automatic identifica-

tion and data capture (AIDC) by integrating drones

with light detection and ranging (LiDAR) devices and

radio-frequency identification (RFID).

Despite the growing interest in adopting drones

for warehouse management tasks, the existing body

of research has primarily focused on diverse domains

like construction, humanitarian efforts, and agricul-

ture. Studies centered on UAV-based warehouse man-

agement exist and are linked with supply chain man-

agement (SCM) and logistics management, [2]. Thus

our research and proposed methodology is essential

for driving UAV adoption in this domain.

The logistics department of industry plays a sig-

nificant role in the global economy today. It is a vi-

tal department of an industry that requires immedi-

ate updates (live- updates) and precision in inventory

monitoring. By succeeding in this, industries would

stay up-to-date and have full knowledge of the prod-

ucts’ sufficiency at any given moment and in a global

status. Industries today rely on their logistics, which

provides constant and reliable inventory information,

to improve their economic value. Therefore, the au-

tomation of the inventory detection and recognition

mechanism is considered necessary to give the real

inventory status of products available to the industry

as quickly as possible and without any human error,

which may occur due to the ”human factor”. By this,

selling goals are succeeded, profit is increased, and

production costs are minimized because ”no unnec-

essary products” are produced.

Using unmanned aerial vehicles (UAVs), com-

monly known as ”Drones,” is believed to improve

this area significantly. Currently, the constant devel-

opment of drones’ vision positioning and GPS inte-

grations are capable of being operated indoors and

outdoors autonomously, without any human interven-

tion, only by giving a flight route plan. It is, there-

fore, crucial to find suitable software that, utilizing

the data provided by the unmanned aircraft, can ac-

curately complete a full inventory scanning by being

able to live record the quantities of products in each

warehouse - branch in any industry’s premises. The

discovery of such an algorithm depends on the chosen

technologies a drone will be equipped with. Depend-

ing on the technologies mentioned above, different

approaches emerge to implementing an efficient algo-

rithm that will complete the above process in the best

possible way. Therefore, careful examination and se-

lection of technological utilities are required, which,

in combination with the algorithm, will produce re-

sults with accuracy and at the lowest implementation

cost.

Regarding the existing research conducted in the

product inventory detection and recognition Field,

[7], the authors focus on developing and evaluating

an aerial robotic system (UAV) for intelligent inven-

torying of accumulated materials within a warehouse.

The LIDAR system collects data points (using PCL

technology). Subsequently, with appropriate post-

processing, the algorithm creates a three-dimensional

model of the environment (the warehouse and stor-

age points) from which material measurements are

taken. Through detailed processing, the researchers

managed to estimate the quantities of bulk materials

in each stack.

In [8], an approximate counting of accumulated

materials (specifically, recyclable waste) located in

an open area is attempted using the process of ”Pho-

togrammetric Surveying.” This process relies on data

processing collected through the photography of the

materials to be measured using an unmanned aerial

vehicle (UAV or Drone).

Furthermore, in [9], research focuses on automat-

ing inventory management in a large product ware-

house. Specifically, they attempt to detect and iden-

tify various products in stock using UAV (Drone)

technology and QRCode technology (Quick Re-

sponse Code).

In [10], the authors propose the use of a high-

resolution portable radio frequency identification

(RFID) reader in a UAV (Drone) for conducting in-

ventory surveys in industrial product warehouses.

Their work aims to find the optimal algorithm for de-

signing the 3D trajectories of unmanned aerial vehi-

cles (UAVs) in complex industrial warehouses, pri-

marily due to the complexity of the UAV process for

the complete detection and reading of all RFID labels

on products within the warehouse proposing the use

of the hybrid fitness-based differential evolution al-

gorithm (PSO-DE).

In [11], the authors propose an approach based

on the use of unmanned aerial vehicles for scan-

ning warehouse inventories and a convolutional neu-

ral network-based deep learning method (R-CNN -

Deep Learning) for autonomous surveying activities.

The research mentioned above cases approach the

problem with relative accuracy and efficiency. How-

WSEAS TRANSACTIONS on SYSTEMS

DOI: 10.37394/23202.2024.23.14

Sotirios Tsakiridis, Apostolos Papakonstantinou,

Alexandros Kapandelis, Paris Mastorocostas,

Alkiviadis Tsimpiris, Dimitrios Varsamis

E-ISSN: 2224-2678

122

Volume 23, 2024

ever, there are several disadvantages to implement-

ing each algorithm. These disadvantages are related

to complex hardware systems (RFID Readers, QR-

Code Readers, and Lidar Scanners), which increase

the implementation cost and affect the flight auton-

omy of drones due to increased total weight. The

above-mentioned research is based on the use of com-

plex algorithms that have a high computational cost,

as well as the use of complex data that are difficult to

manage. As a result, data post-processing is manda-

tory so that the algorithm can complete the process of

recognizing the inventory of the warehouse.

This research aims to find the optimal algorithm

for detecting and recognizing inventory in a ware-

house using an unmanned aerial vehicle (UAV -

Drone), which will capture stereoscopic data using

a LiDAR Scanner. Furthermore, this study tries to

identify the best algorithm that effectively and di-

rectly utilizes the LiDAR Scanner’s stereoscopic data,

unlike the complex data used in existing research.

This allows for completing the inventory detection

and recognition process on the fly without further data

processing. This approach minimizes the cost and

time required for implementing the process, as de-

scribed below.

2 Methodology

In this section, a proposed solution to the above prob-

lem is described. Fig. 1 depicts the basic steps of the

product inventory detection and measurement pro-

cess.

Figure 1: Implementation Steps (Diagram)

2.1 UAV Control and Navigation System

In favor of the best solution to the research, a custom-

made drone constructed from lightweight materi-

als (carbon fiber frame) was chosen. This drone

should be equipped with all navigation control sys-

tems (mainboard, GPS navigation system) and a high-

performance lidar scanner without exceeding its max-

imum payload capacity while maintaining the maxi-

mum possible flight autonomy. This ensures that the

drone can complete the scanning process without fur-

ther interventions. Initially, the drone operator creates

a flight plan by mapping out the predetermined path

the drone will follow using specially designed soft-

ware. The software used, ”Mission Planner”, [12],

utilizes point-to-point navigation technology. On a

map (embedded in the software), the flight’s start-

ing point, route, and endpoint are visually marked.

Each selected point corresponds to specific GPS co-

ordinates. Subsequently, the flight plan is loaded into

the drone’s system, which successfully executes it

through its GPS navigation system. Throughout the

entire process, the operator does not intervene, ex-

cept in cases where there is a safety concern for the

surrounding environment.

In Fig. 2, the actual state of an inventory ware-

house is shown (using a warehouse model for research

purposes).

Figure 2: The representation of an empty warehouse

After receiving the flight plan data through the

process mentioned above, the drone carries out the

flight mission over the warehouse. Specifically, the

drone navigates over the inventory according to the

operator’s pre-defined position points. The operator

predetermines the flight altitude through the program

and is a critical element for the subsequent operation

of the detection algorithm.

2.2 Warehouse Scanning Using LiDAR

Technology

LiDAR technology is utilized for the implementation

of warehouse inventory detection. Specifically, the

drone is equipped with a LiDAR sensor that captures

elevation data. To execute this process, specific sys-

tem parameters must be defined in advance. Initially,

the drone flies directly above the products. The prod-

ucts are originally placed on pallets of exact known

dimensions (length and width of the pallet). Addition-

ally, each pallet has a specific maximum load height,

meaning that each pallet carries a predetermined num-

ber of products. The pallets with products can be

stacked (one pallet on top of another, as this is com-

mon in industrial warehouses) with a specified max-

imum stack height. This parameter is crucial to the

solution given because it determines the flight alti-

tude of the drone. The stacked pallets are placed in

predefined sections of the warehouse (each section is

divided into predefined columns and rows – specific

areas) to ensure that the products on the pallets are

sorted by type and can be detected.

Fig. 3 shows a model of an empty section of a

warehouse. Each section has predefined columns and

WSEAS TRANSACTIONS on SYSTEMS

DOI: 10.37394/23202.2024.23.14

Sotirios Tsakiridis, Apostolos Papakonstantinou,

Alexandros Kapandelis, Paris Mastorocostas,

Alkiviadis Tsimpiris, Dimitrios Varsamis

E-ISSN: 2224-2678

123

Volume 23, 2024

rows where the workers place stacked pallets. Each

“box” represents a pallet or a stack of pallets.

Figure 3: The representation of an empty section of

the warehouse

In Fig. 4, the model represents the exact position

of the drone above every stacked pallet in each prede-

termined area of the warehouse.

Figure 4: The navigation protocol the drone produces

above each section

During the drone’s flight, the LiDAR scanner cap-

tures the distance between the initial lidar position

and the stacked pallets below the drone at regular in-

tervals, which are determined by the software. This

distance represents the metric difference between the

stack’s distance from the drone’s predetermined flight

altitude. The same procedure is applied to each area

of the warehouse until the scanning of all sections of

the warehouse is completed.

2.3 Inventory detection and scanning

algorithm

The data obtained from the LiDAR scanner is ex-

tracted from the drone and is input into the soft-

ware. Subsequently, the algorithm provides an esti-

mation of the number of counted products (in stacks)

in the warehouse. Along with the LiDAR scanner’s

recorded data values, specific parameters necessary

for the algorithm’s correct operation are defined into

the program. Specifically, the program incorporates

the drone’s flight altitude, the LiDAR sampling res-

olution, the predetermined length and width of each

pallet, as well as the specified height that each pal-

let can be loaded. In Fig. 5 and Fig. 6, each dia-

gram illustrates the distance measurements obtained

by the LiDAR scanner in relation to the sampling

frequency. To accurately represent the real state of

the warehouse, the algorithm includes a flight margin

compensation to reduce instances of incorrect mea-

surements of product stacks.

The flight margin compensation in our algorithm

plays a crucial role in mitigating inaccuracies in the

measurement of product stacks during UAV flights.

This compensation accounts for variations in flight

conditions, such as changes in altitude, wind speed,

and other environmental factors, which could impact

the precision of LiDAR measurements. In practical

terms, the algorithm incorporates a margin of adjust-

ment during the flight, allowing for real-time adapta-

tions based on the dynamic conditions encountered.

By doing so, we aim to enhance the accuracy of mea-

surements, ensuring that the LiDAR data collected

aligns more closely with the actual state of the ware-

house at any given moment. This adaptive approach

minimizes the likelihood of errors in the determina-

tion of product stack dimensions, contributing to a

more accurate representation of the warehouse envi-

ronment.

Figure 5: Diagram representing the Lidar’s distance

calculations in Row 1

The above measured distances are converted into

stack height values through a process used by the al-

gorithm to estimate the number of pallets in every

stack of each section in the warehouse.

In Fig. 7, the model represents the results of the

scanning and identification procedure by the algo-

rithm.

To achieve optimal results, the algorithm groups

the values obtained from the LiDAR scanner into val-

ues of the same or similar height, considering the pre-

defined error margin. Each subsequent value is in-

cluded within the existing value group if it falls within

the bounds of the mean value of the current value

group, based on the specified error margin. Each

value group corresponds to the width of each stack

of pallets located on the same line in each warehouse

area, only in case they have the same value as the

mean of the current value group. In cases where val-

WSEAS TRANSACTIONS on SYSTEMS

DOI: 10.37394/23202.2024.23.14

Sotirios Tsakiridis, Apostolos Papakonstantinou,

Alexandros Kapandelis, Paris Mastorocostas,

Alkiviadis Tsimpiris, Dimitrios Varsamis

E-ISSN: 2224-2678

124

Volume 23, 2024

Figure 6: Diagram representing the Lidar’s distance

calculations in Row 2

Figure 7: Results of the inventory scanning procedure

ues differ from the mean, a new stack is created. The

differentiation of consecutive stacks of pallets of the

same height is achieved by comparing the number of

values in the group (along with the error margin) with

the predetermined width of each pallet. The LiDAR

sampling resolution is also taken into account as a pa-

rameter. This process ensures that the algorithm will

group lidar measurements and recognize each consec-

utive stack of pallets without the risk of incorrect esti-

mation, due to the fact that some stacks of pallets may

have the same height in each measuring line.

3 Results ans Discussion

The above procedure was tested and evaluated using

random given data in a model. The results were evalu-

ated at 100%. The same process is used to fully scan

and estimate the exact inventory of each section in

a warehouse. The same navigation protocol is used,

specifically modified for more sections in the ware-

house. In Fig. 8, the navigation process is being rep-

resented.

In Fig. 9 and Fig. 10, each diagram illustrates the

distance measurements obtained by the LiDAR scan-

ner in relation to the sampling frequency.

Obtained data from the drone are inserted into the

program. The results of the algorithm are evaluated at

100%. In Fig. 11, we represent the evaluation results

of a total warehouse scanning.

Figure 8: Navigation guidance through every section

of a warehouse

Figure 9: Scanning Procedure for Rows 1 and 2

Figure 10: Scanning Procedure for Rows 3 and 4

Figure 11: Results of Warehouse Inventory estima-

tion

WSEAS TRANSACTIONS on SYSTEMS

DOI: 10.37394/23202.2024.23.14

Sotirios Tsakiridis, Apostolos Papakonstantinou,

Alexandros Kapandelis, Paris Mastorocostas,

Alkiviadis Tsimpiris, Dimitrios Varsamis

E-ISSN: 2224-2678

125

Volume 23, 2024

Unmanned Aerial Vehicles (UAVs) have emerged

as game-changers in the realm of industrial inventory

management. Leveraging state-of-the-art technology,

UAVs equipped with LiDAR sensors provide an inno-

vative solution for the detection and quantification of

inventory in large-scale warehouses. This approach

reduces reliance on traditional manual methods, en-

hancing efficiency and accuracy in assessing stock

levels. The integration of UAVs introduces a dynamic

and adaptable system that autonomously navigates

warehouse spaces, capturing detailed LiDAR data to

precisely quantify inventory items. This transforma-

tive technology holds the promise of revolutionizing

how industrial warehouses manage and optimize their

inventory processes.

In the pursuit of advancing warehouse logistics,

the implementation of UAV-based inventory detec-

tion offers a paradigm shift. This approach not

only streamlines the traditional inventory manage-

ment process but also introduces a cost-effective and

time-efficient solution for industries grappling with

the challenges of large-scale warehouse operations.

As industries increasingly recognize the potential of

UAVs in this domain, the integration of LiDAR tech-

nology for accurate and real-time inventory assess-

ments becomes a pivotal advancement, marking a sig-

nificant step toward the future of smart and automated

warehouse management systems.

4 Conclusion

In subsequent stages, the Goal is to implement the de-

tection of inventory within a warehouse with pallets

and products placed in unspecified positions. This

hypothesis represents the primary objective of the re-

search, as in all industrial warehouses, predetermined

product placement positions cannot exist, because ev-

ery section is continuously modified to serve various

types of product placement. Furthermore, the future

aim of this research is to complete the entire inven-

tory detection and inventory scanning process, using

the existing resources and achieving on-the-fly detec-

tion. We are encouraged to complete the above whole

process using special hardware and software attached

to the drone, as to achieve real-time monitoring of an

industrial warehouse.

Acknowledgment:

This research work was carried out as part of the

project ”Optimization of placement and counting

products in large industrial areas using UAV”

(Project code: KPM6-0083129) under the

framework of the Action ”Investment Plans of

Innovation” of the Operational Program ”Central

Macedonia 2014 2020”, that is co-funded by the

European Regional Development Fund and Greece.

References:

[1] F. P. Mahmoud Almasri, Xavier Marjou,

“Reinforcement-learning based handover

optimization for cellular uavs connectivity,”

WSEAS Transactions on Computer Research,

vol. 10, pp. 93–98, 2022.

[2] A. Fotia, R. Pucinotti, and V. Barrile,

“Detection of steel structures degradation

through a uavs and artificial intelligence

automated system,” WSEAS TRANSACTIONS

ON CIRCUITS AND SYSTEMS, vol. 21,

pp. 231–237, 12 2022.

[3] C. Malang, P. Charoenkwan, and

R. Wudhikarn, “Implementation and critical

factors of unmanned aerial vehicle (uav) in

warehouse management: A systematic

literature review,” Drones, vol. 7, p. 80, 2023.

[4] H. Cho, D. Kim, J. Park, K. Roh, and

W. Hwang, “2d barcode detection using images

for drone-assisted inventory management,”

2018.

[5] I. Kalinov, A. Petrovsky, V. Ilin, E. Pristanskiy,

M. Kurenkov, V. Ramzhaev, I. Idrisov, and

D. Tsetserukou, “Warevision: Cnn barcode

detection-based uav trajectory optimization for

autonomous warehouse stocktaking,” IEEE

Robotics and Automation Letters, vol. 5, 2020.

[6] M. Beul, D. Droeschel, M. Nieuwenhuisen,

J. Quenzel, S. Houben, and S. Behnke, “Fast

autonomous flight in warehouses for inventory

applications,” IEEE Robotics and Automation

Letters, vol. 3, 2018.

[7] R. M. Gago, M. Y. A. Pereira, and G. A. S.

Pereira, “An aerial robotic system for inventory

of stockpile warehouses,” Engineering Reports,

vol. 3, no. 9, p. e12396, 2021.

[8] G. Tucci, A. Gebbia, A. Conti, L. Fiorini, and

C. Lubello, “Monitoring and computation of

the volumes of stockpiles of bulk material by

means of uav photogrammetric surveying,”

Remote Sensing, vol. 11, no. 12, 2019.

[9] B. Yoon, H. Kim, G. Youn, and J. Rhee, “3d

position estimation of objects for inventory

management automation using drones,”

Applied Sciences, vol. 13, no. 19, 2023.

[10] Y. Han, Q. Chen, N. Pan, X. Guo, and Y. An,

“Unmanned aerial vehicle 3d trajectory

planning based on background of complex

industrial product warehouse inventory,”

Sensors and Materials, vol. 34, p. 3255, 08

2022.

WSEAS TRANSACTIONS on SYSTEMS

DOI: 10.37394/23202.2024.23.14

Sotirios Tsakiridis, Apostolos Papakonstantinou,

Alexandros Kapandelis, Paris Mastorocostas,

Alkiviadis Tsimpiris, Dimitrios Varsamis

E-ISSN: 2224-2678

126

Volume 23, 2024

[11] A. De Falco, F. Narducci, and A. Petrosino,

“An uav autonomous warehouse inventorying

by deep learning,” in Image Analysis and

Processing – ICIAP 2019: 20th International

Conference, Trento, Italy, September 9–13,

2019, Proceedings, Part I, (Berlin, Heidelberg),

p. 443–453, Springer-Verlag, 2019.

[12] ArduPilot, “Mission planner.” Accessed on

December 10, 2023.

Contribution of Individual Authors to the

Creation of a Scientific Article (Ghostwriting

Policy)

The authors equally contributed in the present re-

search, at all stages from the formulation of the prob-

lem to the final findings and solution.

Conflicts of Interest

The authors have no conflicts of interest to

declare that are relevant to the content of this

article.

Creative Commons Attribution License 4.0

(Attribution 4.0 International , CC BY 4.0)

This article is published under the terms of the

Creative Commons Attribution License 4.0

https://creativecommons.org/licenses/by/4.0/deed.en

_US

WSEAS TRANSACTIONS on SYSTEMS

DOI: 10.37394/23202.2024.23.14

Sotirios Tsakiridis, Apostolos Papakonstantinou,

Alexandros Kapandelis, Paris Mastorocostas,

Alkiviadis Tsimpiris, Dimitrios Varsamis

E-ISSN: 2224-2678

127

Volume 23, 2024

Sources of Funding for Research Presented in a

Scientific Article or Scientific Article Itself

No funding was received for conducting this study.