Hindawi

Journal of Robotics
Volume 2023, Article ID 9967236, 17 pages
https://doi.org/10.1155/2023/9967236

Research Article
Real-Time SLAM Mobile Robot and Navigation Based on
Cloud-Based Implementation

1,2 1
Jaafar Ahmed Abdulsaheb and Dheyaa Jasim Kadhim
1
Department of Electrical Engineering, College of Engineering, University of Baghdad, Baghdad, Iraq
2
Department of Electronics and Communication, College of Engineering, Uruk University, Baghdad, Iraq

Correspondence should be addressed to Jaafar Ahmed Abdulsaheb; jaafer@uruk.edu.iq

Received 3 January 2023; Revised 10 March 2023; Accepted 15 March 2023; Published 29 March 2023

Academic Editor: L. Fortuna

Copyright © 2023 Jaafar Ahmed Abdulsaheb and Dheyaa Jasim Kadhim. Tis is an open access article distributed under the
Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided
the original work is properly cited.
Tis study investigates the feasibility of a mobile robot navigating and discovering its location in unknown environments, followed
by the creation of maps of these navigated environments for future use. First, a real mobile robot named TurtleBot3 Burger was
used to achieve the simultaneous localization and mapping (SLAM) technique for a complex environment with 12 obstacles of
diferent sizes based on the Rviz library, which is built on the robot operating system (ROS) booted in Linux. It is possible to
control the robot and perform this process remotely by using an Amazon Elastic Compute Cloud (Amazon EC2) instance service.
Ten, the map to the Amazon Simple Storage Service (Amazon S3) cloud was uploaded. Tis provides a database to display maps
and use them at any time for navigation without the need to redraw the map. Tis map can be accessed by using an authentication
process (username and password) supervised by the cloud server administrator. After that, using the serverless image handler
(SIH), with the aid of this solution, you can change the size of images, change the color of the background, format them, or add
watermarks. Experiment results demonstrated the ability to build a map of an unknown location in a complex environment and
use it for navigation tasks on a real mobile robot via remote control. It also showed the success of the process of storing the map for
future use and the process of modifying the map using SIH.

1. Introduction proposed the creation of a cloud-connected robotics net-

work that uses fexible resources when requested from the
Due to the growing need for robots in our lives, the robotics global cloud architecture [3]. Cloud computing philosophies
industry has seen many important changes, such as auto- have been applied by researchers in robotics as well as
mated production, extraterrestrial missions, unmanned mobile environments, resulting in topics such as cloud
rescue vehicles, disaster rescue operations, socially sup- robots and mobile clouds. When connected to the cloud,
ported robots at home and in education, self-driving ve- these robots can take advantage of the powerful computing
hicles, and therapeutic robots used in health, Mars survey, or and communications resources in the cloud for the data
seabed analysis [1]. One of these important developments in center and storage and can also process and share in-
the feld of robots is the use of networked robots to share formation with many robots [4]. Cloud Computing (CC)
information and provide required resources such as data can be used to upgrade the capabilities of robots in many
storage, complex computation, and so on. Tese networked tasks since cloud computing features are increasingly useful
robots can get these information and resources through for mobile robots. In general, there are two types of mobile
servers that are connected to the Internet and programs that robots: autonomous mobile robots, which can explore their
they have installed [2]. To address the proposed difculties environments on their own, whether they are known or
that the robots may face in accessing the information and unknown, and nonautonomous mobile robots, which may
required resources, developers and researchers have recently use a guidance system to move through their surroundings
2 Journal of Robotics

[5]. Tis work deals with autonomous mobile robots environments that are divided into subdivided maps. Also,
(AMRs), which are needed in many felds to operate in- the whole topological demonstration and necessary in-
dependently of humans. Planning a better path, while formation of the world are saved in cloud structures from
avoiding obstacles, is essential for AMR to move from one a distance by employing fxed ecological tags that consist of
point to another, representing the target in their environ- collections of QR codes as well as AR tags. Tests are being
ment. Tere are several implementations that deal with the done to determine the possibility of selection, adjust the
feld of robotics, such as SLAM, which is abbreviated for sizes, and see if the proposed procedure can be done.
simultaneous localization and mapping, as well as the large Megalingam et al. [11] presented the ROS implementation of
number of sensing and grasping systems that quickly lead SLAM by calculating how long the robot model would take
towards the emergence of large amounts of sensor data. to arrive. Te test is conducted in an Rviz-created virtual
Tese data are difcult to store because most robots have environment. Te travel time is calculated by placing various
limited storage capacity. Cloud robotics (CR) submits so- dynamic obstacles for various destinations on the map. It has
lutions towards saving space within the cloud so that all- been seen that the robot has a good response time and travels
important robot information can be stored inside this cloud from the source to the destination in a reasonable amount of
for future use. Tis feature allows the CR to authorize mobile time, where the amount of time needed increases as the
robotics to access huge datasets such as general localization distance does. Jiang et al. [12] presented SLAM, which was
maps and platforms of the environment [6]. put into practice by using Turtlebot2, a mobile robot, and the
Cloud robotics has become increasingly popular in re- robot operating system’s navigational package. Te Tur-
cent years. Many works have a lot to do with the ways to do tlebot2 is outftted with a notebook, a 3D-LiDAR ranging
research that are talked about in this section. Te study by sensor, an RGB-D camera, and light detection. Tese fea-
Afanasyev et al. [7] presented the SLAM algorithm based on tures are used to help the robot navigate and create a full 3D
ROS for the simulated mobile robot on the Gazebo simu- sight map in addition to a novel 2.5-dimensional map. Te
lator, which is moveable in a three-dimensional pattern of conclusion of the collected results displays a collection of
the actual internal situation that is introduced and is ofered image and scans data able to enhance the performance of
in a 3D model form based on the actual interior environ- mapping using classical approaches. Te study by Arena
ment. Ten, the simulation results of the willow garage et al. [13] describes an innovative robot that looks like
personal robot (PR2) robot are described. Te camera shots a worm and is made of an ionic polymer-metal composite
taken with autodesk’s Rao-Blackwellized Particle Filter (IPMC) self-actuated skeleton. It is controlled by cellular
(RBPF) and the data from the PR2 robot’s laser for locali- neural networks (CNNs). Worm locomotion is performed
zation and mapping are used to create an image-based 3D by bending the actuators sequentially from “tail” to “head,”
model of the actual area. In the study by Doriya et al. [8], imitating the traveling wave observed in real-world un-
diferent cloud architectures and how they connect to one dulatory locomotion. Te study by Buscarino et al. [14] looks
another or difer from one another in their ability to solve into and suggests a way to control a group of mobile robots
SLAM problems are discussed and compared. So, every without a central point of control using a dynamical network
framework uses various collections of special protocols model. Te results show that the performance of the system
within the cloud that enable the mobile robot to delegate its can be improved by letting the robots talk to each other over
computational tasks to the cloud. Te researchers confrmed long distances, and a good way to control exploration and
that the framework of Rapyuta is more elastic than other transport is presented. Small-world networks can be highly
frameworks, and because its implementation is open source, clustered, like regular lattices, yet have small characteristic
it can be expanded to include functions of other robots by path lengths, making them “small-world” dynamical systems
several other researchers. Whereas, when they compared it with enhanced signal-propagation speed, computational
to DaVinci, they found that DaVinci was the most suitable power, and synchronizability [15]. Te study by Kajita and
for SLAM. Finally, the cloud framework for cooperative Espiau [16] talks about the history of research on two-legged
tracking and mapping (C2TAM) framework is suitable for robots, the dynamics and control of general two-legged
saving huge amounts of data. robots, the zero-moment point (ZMP), the relationship
Takaya et al. [9] introduced a simulation environment between gaits and stability, and how two-legged robots are
for mobile robots based on ROS and Gazebo. Te code diferent from each other. It also talks about performance
advanced for the simulation procedure can be directly ex- indices like the Froude number and the specifc resistance, as
ecuted on the actual robot without modifcations after well as how things will change in the future. Te authors in
properly creating the models of the robot under the gazebo [17] made a web interface that lets you control and talk to
simulator. Ten, the tasks of autonomous navigation and multiple robots using ROS topic identifcation and regis-
3D-mapping simulation utilizing control programs under tration for autonomous robots. Te Gazebo simulator sends
ROS are ofered. Te results of the simulation and experi- out all robots to interact with a user. To test how well the
ments coincided very well and showed the usability of the system works, the number of robots was increased. Te big O
advanced environment. For this study, the ROS platform representation was used to analyze the running-time
was used to build models of the PeopleBot and Pioneer 3-DX complexity of algorithms. Te experiment result indicated
robots. Limosani et al. [10] proposed a system based on the that the autonomous robot registration was successful and
cloud robotics model that lets automated mobile robots that the communication performance decreased gradually.
move through and observe previously unknown interior Te study by Rashid et al. [18] introduces a new algorithm,
Journal of Robotics 3

called cluster matching, for multirobot localization and 2. Research Methods and Tools
orientation. It involves a distance IR sensor scanning the
robots and estimating the absolute positions and orienta- 2.1. Navigation and SLAM. Te navigation of robotics de-
tions of a number of the team robots without knowing their notes the activity of a robot to control its location in relation
IDs. Te localization and orientation of robots not visible to to a specifc reference and then outline a track to arrive at the
the distance IR sensor are obtained by collecting the in- wanted location. It can contain a set of jobs such as track
formation coming from the on-board sensors and recon- planning, localization, and mapping [19]. Te procedure for
structing a complete map of the team distribution. building a map demands large amounts of storage space and
Simulation scenarios are implemented on tens of robots to is computationally heavy. Moreover, searching for a map
show the performance of the algorithm. demands large quantities of information and data, and this is
Te main problems have been considered and are our a huge challenge, especially when the region of navigation is
motivation to move forward with the suggested oversight large. CR delivers an actual, favorable solution for the up-
strategies. Tese problems are listed sequentially as follows: coming cloud based on navigation. Also, the cloud not only
supplies storage area to save a large quantity of map data but,
(1) Navigation is the frst and main problem for a mobile moreover, supplies processing strength to make it easy to
robot that needs to navigate its unknown environ- quickly build and search the map [20]. A prerequisite for
ment and defne its location so that it can design successfully navigating is the ability to localize and create
a path to other destinations. SLAM technology is maps for unknown environments, which allows the robot to
considered the dominant solution to such problems. navigate in them using SLAM.
(2) Te other challenge with SLAM technology occurs SLAM was frst presented in mid-1986 by Smith and
during navigation in unknown environments since Durrant-Whyte [21]. SLAM is considered one of the AI
the robot takes a long time to explore the whole techniques that benefts from the mobile robot’s navi-
environment to build its map. When another robot gation in an unknown environment to construct a new
has to navigate the same environment, it will rebuild map for this environment. When a mobile robot is placed
a new map. So, we need to save all constructed maps, in an unknown environment, it has to move and acquire
while the storage capacity needs to store the acquired data about landmarks in the environment by using its
data, such as maps, within private or public clouds, sensors, while at the same time, it has to assign its po-
so that it can access them at any time and from sition concerning that created map. Te landmark is
anywhere via the Internet. defned as a symmetrical object that can be easily dis-
Te principal contributions made by this research tinguished from a scanning sensor, and it is utilized to
project are listed as follows: update the evaluated position of the robot [22]. Terefore,
a good identifcation based on good benchmarks is im-
(1) Simulating and implementing the SLAM algorithm portant to support the operation of SLAM, as it cannot
in an unknown environment that is built on the Rviz rely only on the frequently inaccurate odometry ap-
simulator on the ROS. Following that, when the proach, which is the most common technique that is
number of obstacles reaches 12, study and evaluate utilized in robotics to evaluate its present position. Te
a mobile robot platform named TurtleBot3 Burger in odometer function integrates growing information as the
the Rviz simulator and implement the SLAM algo- motion of the wheels relative to the starting position is
rithm for diferent environments based on the Rviz recorded. So, the SLAM problem can be defned as a case
library, which is built on the ROS in the Linux study of a chicken-or-egg problem, since building a map
operating system. Also, control the robot and per- of the environment needs the robot position as well as
form this process remotely. assigning the robot position needs the map of the en-
(2) Simulating an Amazon Web Service (AWS) cloud vironment. Te main problem with SLAM is that the
server to store maps made by robots and to store and measurements that are read from the sensors are always
download diferent required fles (PDF, words, video, noisy and their positions are not always certain because
and photos), which can be accessed through an the robot is moving.
authentication process (username and password)
overseen by the cloud server administrator.
2.2. TurtleBot3 Burger. Te TurtleBot3 Burger is a small
In this study, SLAM is implemented by using a mobile mobile robot that runs on ROS and can be programmed
robot named Turtlebot3 Burger, using gmapping tools in [23]. Te TurtleBot3 burger is one of two robots, with the
ROS based on the Rviz platform, and by placing diferent TurtleBot3 wafe being the other. Te wafe is lower, but it is
objects on the map and in diferent scenarios. Tis article is wider all around. Te robot’s base can move around, thanks
organized as follows: Section 2 discusses navigation and to two wheels and a metal ball that keeps it steady. It can be
simultaneous localization and mapping (SLAM) and de- fully programmed with open source, and the software that
scribes hardware and software tools; Section 3 discusses comes with it is licensed under an Apache 2.0 license.
Modeling of Mobile Robot Navigation, and Section 4 dis- Figure 1 shows the three diferent TurtleBot3 versions, and
plays and discusses the simulation results. Finally, the Figure 2 shows the TurtleBot3 Burger, while Table 1 shows
conclusion of this paper is represented in Section 5. the specifcations of the TurtleBot3 Burger.
4 Journal of Robotics

Table 1: Specifcation of the TurtleBot3 Burger.

Original TurtleBot TurtleBot3
(Discontinued) Items
Burger specifcation
Maximum translational
TurtleBot 2 Family TurtleBot 3 Family 0.22 m/s
Burger velocity
Maximum rotational velocity 2.84 rad/s (162.72 deg/s)
Maximum payload 15 kg
Size (L × W × H) 138 mm × 178 mm × 192 mm
TurtleBot 2 TurtleBot 2i Wafe Wafe Pi
Weight
1 kg
(+SBC + battery + sensors)
SBC (single board computers) Raspberry Pi
TurtleBot 2e TurtleBot Euclid
Embedded controller
M7)
®
OpenCR (32 bit ARM Cortex -
®
Figure 1: Turtlebot family.
HLS-LFCD2
3-axis gyroscope
Sensor
360° LiDAR for SLAM & Navigation
3-axis accelerometer
3-axis magnetometer
Scalable Structure

3. Modeling of Mobile Robot Navigation

Single Board Computer (Raspberry pi)
In this part, the Rviz platform is proposed to use SLAM
OpenCR (ARM Cortex-M7) technology to simulate how mobile robots move through an
DYNAMIXEL x 2 for Wheel
unknown environment, control the robot, and do all of these
remotely. For practical considerations, a mobile robot plat-
Sprocket wheels for Tire and Caterpillar form named TurtleBot3 Burger has been adopted here to be
Li-Po Battery 11.1 V 1,800 mAh used to do tests in the Rviz library built on ROS on Linux. Te
procedure begins with navigating in an unknown indoor
Figure 2: Te TurtleBot3 Burger. environment with many obstacles, which it must detect, and
creating a map of this unknown environment. As a result, the
2.3. Robot Operating System (ROS). ROS is an operating SLAM algorithm accomplishes all of these goals and can be
system for robots that includes drivers, libraries, and tools controlled via cloud services access, also known as AWS.
that can be used to build and improve robot systems [24]. Figure 3 shows the software architecture that came out of this.
Willow Garage made it in 2007, and it is now the most Te program is run on a master PC, which the robot acts as
popular and widely used robotics platform. ROS has Linux a slave to. A “roscore,” which is the central ROS process that
as a command instrument and an interprocess communi- connects nodes, needs to be started by the hand by running
cation system [25]. It contains many autonomous nodes, a command on the RPi through an SSH connection. A similar
each of which communicates with the other nodes by uti- process is started in the background on the master PC by the
lizing publish or subscribe messaging patterns. ROS relies on robotics system toolbox. After the two processes have been set
various languages, such as Python and C++, which can be up, they share data that can be used on either the PC or the
utilized with each other in the ROS system, which runs on RPi. Te TurtleBot 3 and the remote computer must be
the Ubuntu Linux operating system [26]. confgured to set up communications on the same network.
Control of the remote computer and storage space is required
to store acquired data, such as maps, in private or public
2.4. Rviz Platform. In computer science and computer clouds, so that it can access them at any time and from
imaging, it is necessary to describe and visualize information anywhere via the Internet. Te procedure for SLAM tech-
in the real world. To achieve this purpose, several structures niques in ROS with AWS is shown in Figure 4.
and diferent types of algorithms are proposed for data
imaging world. Unfortunately, often the data stream plat-
form that is known to handle the issues in visualization 3.1. Cloud Computing. IT departments and developers can
systems is not elastic and is not suitable to imagine newly focus on what matters most by using cloud computing
invented data frames and algorithms since this sketch can instead of doing undiferentiated tasks such as capacity
only admit specifed data frameworks. So, to handle this planning, maintenance, and buying. A variety of models and
issue, a novel instrument of visualization that is defned by deployment strategies have emerged as cloud computing has
Rviz is suggested [27]. Rviz is defned as a 3D imagining grown in popularity, in order to meet the various needs of
instrument in ROS that allows us to imagine what the mobile users. Depending on the type of cloud service you use and
robot will grasp and see in any area. Rviz can also imagine all the deployment strategy, you have varying degrees of
of the types of sensors on the robot, which include laser control, fexibility, and management. Knowing the difer-
scanners, cameras, and point clouds. Observing the raw ences between infrastructure as a service (IaaS), platform as
values of the sensors [28], Rviz has an autonomous view of a service (PaaS), and software as a service (SaaS), as well as
the input data structures. what deployment methodologies you can use, will help you
Journal of Robotics 5

Amazon Web Master/ Slave/

Services Remote PC access 24/7 Remote PC Turtlebot
wherever safe and quick. SSH
Windows/
EC2 Linux
Linux

Download/Upload
Data WIFI
S3 ROS ROS

Robotics
SIH System
Toolbox
Optional

TB_tools

(a) (b) (c)

Figure 3: Software architecture. (a) AWS services are used to remotely control (services explained in the next subsection 3.2). (b) A remote
PC is used to control TurtleBot3. (c) Te TurtleBot3 used for SLAM and navigation techniques.

the provider manages the infrastructure. SaaS ofers ready-

User Installing ROS (PC setup) to-use and out-of-the-box solutions to meet specifc business
needs. Figure 5 illustrates the distinctions between IaaS,
PaaS, and SaaS. In this work, AWS is a provider of IaaS. As
Creating an AWS account
Single Board Computers the most complete and widely used cloud platform in the
(SBC) Setup
world, AWS provides over 200 fully functional services from
data centers around the world. Millions of clients use AWS
to reduce costs, increase agility, and accelerate innovation,
Launch EC2 instance OpenCR Setup
including the largest corporations, most successful gov-
ernmental organizations, and fastest-growing start-ups.
Compared to other cloud providers, AWS ofers a much
Choose Number of instances Bringup TurtleBot3 wider range of services and features within those services. As
a result, moving your current applications to the cloud is
quicker, simpler, and more cost-efective, and you can build
Enable
auto-assignment of public IP Teleoperation
almost anything you can think of. In the next subsection,
addresses some of these services are explained.

Confgure security group SLAM 3.2. Cloud Services Access (AWS). Te collection of cloud
services ofered by Amazon.com, Inc. is known as AWS
services. It started out in July 2002 by ofering storage and
Create new key pair Save Map
computing services. Since then, it has grown into a fexible
cloud platform with IaaS, PaaS, and SaaS services, as well as
a number of other services.
Create an S3 bucket, upload AWS is dispersed among numerous datacenters located
Connect to windows instance
a map to the S3 bucket in various cities around the globe. Every location is a region,
and each region is made up of physically distinct datacenters
Access any device from a called availability zones. Each service provided distinct
Windows instance. Whenever
Serverless Image Handler availability zones and regions, with some limited to a single
you are, quick and safe.
availability zone and others spanning multiple regions. Some
services also provide replication capabilities across regions
and availability zones. Te following is an explanation of the
Navigation diferent AWS services that are important to this study:
(1) Te simple storage service, also known as S3, is
Figure 4: SLAM technique procedure in ROS.
Amazon’s cloud-based storage service. It allows users
to store standard fles (objects) up to 5 TB in size and
choose the set of services that are best for your needs [22]. As only charges for the storage space actually used. It
far as the features they ofer, IaaS, PaaS, and SaaS are dif- uses buckets, each of which may hold an infnite
ferent from each other in the following ways: IaaS gives the number of things and has a name that is globally
most freedom when it comes to hosting custom-built apps unique [29]. A bucket is located in a region and that
and storing data in a general data center. PaaS allows de- region contains numerous places for storing all of its
velopers to spend more time developing applications while data. Trough API requests and the management
6 Journal of Robotics

On Premises IAAS PAAS SAAS

(own server) (virtual machines) (app service) (0365)

Applications Applications Applications Applications

Data Data Data Data

Runtime Runtime Runtime Runtime

Middleware Middleware Middleware Middleware

O/S O/S O/S O/S

Virualization Virualization Virualization Virualization

Servers Servers Servers Servers

Storage Storage Storage Storage

Networking Networking Networking Networking

You manage Manage by cloud service provider

Figure 5: IaaS, PaaS, and SaaS.

console, data can be moved to another region. A key

that mimics a fle path on ∗ NIX systems is used to
uniquely identify each item in a bucket. Tey are all
accompanied by metadata, and each one can have
Up/Download
a diferent set of rights applied to it. An object may be User
an Object
Internet AWS S3
made publicly accessible, after which it may be
downloaded using an HTTP URL. S3 can be accessed Figure 6: AWS S3 design.
through the AWS management web interface, SOAP,
and REST APIs, and access can be controlled with storage (EBS) volumes can employ up to 4,000 preallocated
the help of access keys and certifcates. As Figure 6 input/output operations per second (IOPS) and have
a maximum capacity of 1 TB. Te cost of volumes is de-
shows, the Management Console, the command-line
interface (CLI), SDKs, and third-party tools can all termined by their size and for provisioned IOPS, by the
upload and download objects to S3 over HTTPS. number of hours allotted to them. EC2 instances can be set
to be EBS optimized when they are launched, improving
(2) Elastic Compute Cloud, or EC2, is the name of performance. A key pair (to verify your identity) and a se-
Amazon’s IaaS product. It was introduced in 2006 curity group are specifed when launching your instance to
and provides virtual machines with various specs secure it. Te private key of the key pair you specifed when
across all the AWS availability zones [30]. An AMI, launching your instance must be provided when connecting
or amazon machine image, is a customized operating to it, as shown in Figure 7.
system that is installed on an EC2 virtual instance. It
is possible to build customized AMIs by either al-
tering an existing image or by developing a brand 3.3. Serverless Image Handler. With the aid of this solution,
new one if any of the available ofcial or unofcial cost-efective image processing can be started in the AWS
AMIs does not satisfy the user’s demands. Custom cloud using a serverless architecture. Te architecture is
AMIs can be used privately, in which case they can designed for dynamic image manipulation and combines
hold crucial data like authentication keys for S3 AWS services with sharp open-source image processing
access, or they can be made available to the public. software and maintains high-quality images with the help of
the solution’s dynamic image handling. Figure 8 shows how
When EC2 frst came out in 2006, the only storage to set up the serverless architecture.
options for instances were S3 for long-term storage and
ephemeral instance storage for fles that only needed to be 4. Experimental and Simulation Results
kept for a short time. Elastic block storage, amazon’s storage
area network (SAN) solution for EC2, was introduced in Tis section shows the results of the implementation of
2008, which provides EC2 instances with access to net- SLAM in ROS, which was simulated in Rviz using an AWS
worked block devices for permanent storage; elastic block EC2 instance service for real-time SLAM mobile robots in
Journal of Robotics 7

Internet

Instance
Security Group

00
Root Private key
Amazon EBS

Availability Zone

Figure 7: AWS EC2 frewall [31].

AWS Cloud

Client Amazon S3 bucket

Amazon Amazon API AWS (customer created)
CloudFront Gateway Lambda

Optimized image 30 KB Original image

Same width and height 100 KB

Figure 8: SIH architecture [32].

Table 2: EC2 server specifcations.

Software image-amazon machine image (AMI) Microsoft Windows server 2022 base with containers
Virtual server type (instance type) t2.micro
Firewall (security group) Create security group
1 volume(s), 30 GiB
Storage (volumes)
General purpose SSD (GP2)
Allow remote desktop protocol (RDP) trafc from Anywhere with user and password
Cost Free tier eligible

complex environments with 12 obstacles of diferent sizes. two levels of security. Te frst is the EC2 user and password,
Tis allows the robot to be controlled from anywhere, and and the second is the user and password for the remote PC
the map is then uploaded to the S3 cloud. Tis provides that controls the TurtleBot3. Te real TurtleBot3 Burger
a database to display maps and use them at any time for robot then starts drawing the map of the unknown envi-
navigation without the need to redraw the map. It also ronment. Figure 10 shows how the EC2 server is monitored
maintains the privacy of the maps in the place in which they while the remote computer is in charge of keeping its re-
are saved, fnally bringing up the map for navigation. All the liability, availability, and performance at a high level. Fig-
detailed steps can be shown in Appendix A: testing slam ure 11 depicts the Real-time SLAM TurtleBot3 control.
techniques, including ros installation, single board computer Figure 12 compares the real environment with twelve ob-
(SBC) setup, OpenCR setup, bringup, teleoperation slam, stacles of diferent dimensions and the environment that is
and navigation; and Appendix B: obstacle detection and drawn by the Rviz platform. Figure 13 shows the upload of
point operation, while EC2 server specifcations can be the environment to AWS s3 and the SIH solution. Figure 14
shown in Table 2. Tis process can be abbreviated as shown shows the navigation process.
in Figure 9. CPU utilization shows how many of the instance’s EC2
First, three levels of security are applied. Te frst is to compute units are being used at any given time; network in
sign in to the AWS account. Ten, the EC2 server connects shows how much incoming network trafc is coming into
to the remote computer and takes control of it with the other the instance; network out shows how much network trafc
8 Journal of Robotics

Sign in to the Command the Upload the Download the

console AWS EC2 real Turtlebot3 map map
User SLAM AWS S3 Navigation
Access at any INSTANCE
time
From anywhere
Always secure
SIH

Figure 9: Scenario process.

CPU utilization (%) Status check failed (any) (count) Status check failed (instance) (count) Status check failed (system) (count)
Percent Count Count Count
26 1 1 1

13 0.5 0.5 0.5

0 0 0 0
21:15 21:30 21:45 22:00 22:15 21:15 21:30 21:45 22:00 22:15 21:15 21:30 21:45 22:00 22:15 21:15 21:30 21:45 22:00 22:15

Network in (bytes) Network out (bytes) Network packets in (count) Network packets out (count)
Bytes Bytes Count Count
7.97 M 9.26 M 12.5 k 12.2 k

3.99 M 4.63 M 6.25 k 6.11 k

0 0 0 0
21:15 21:30 21:45 22:00 22:15 21:15 21:30 21:45 22:00 22:15 21:15 21:30 21:45 22:00 22:15 21:15 21:30 21:45 22:00 22:15

Figure 10: EC2 monitoring.

(a)

(b)

Figure 11: Via AWS EC2: (a) the start of the slam; (b) end of slam.
Journal of Robotics 9

Figure 12: Rviz platform environment vs. real environment.

(a)

(b)

Figure 13: (a) Upload of the environment to AWS S3; (b) the SIH solution.

leaves a particular instance; and status check failed reports In Figure 11, it can be noted by the visual comparison
whether the instance has recently passed the system status that the arena drawn by Rviz is identical to the “real feld,”
check. Table 3 is displaying the Amazon EC2 Monitor. where the robot was able to discover all the obstacles of
Figure 10 shows that the robot is fully controlled by diferent sizes. It is now possible to save the map in AWS S3
AWS EC2. Tis shows the beginning of the process of and use it in the future from the same robot or any other
drawing the map of an unknown place and the end of the robot without having to redraw it by redownloading it from
process. Tis method can allow us to control the robot even AWS S3 and getting the most out of it. Te same principle
if it is in one country and the person in control is in another can be applied to a group of sites, and these maps can be used
country. Te properties of the robot are mentioned in to make a real application that serves those sites, like a large
subsection 2.2. It is possible to control the speed of the store or military application, or to help fnd survivors after
robot so that the maximum speed is 0.22 m/s. Tis is the a disaster.
relationship between the speed of the drawing and the Figure 13 shows the process of uploading the fle in pgm
speed of the robot. format, which is slam, and png format, which is the image
10 Journal of Robotics

(a) (b)

(c) (d)

(e) (f)

Figure 14: Navigation process.

Table 3: Amazon EC2 monitor.

Max value after one
Metric
hour
CPU utilization 26 percentage
Network in 8 MB
Network out 9.26 MB
Network packets in 12.5 k count
Network packets out 12.2 k count
Status check failed 0
Status check failed (instance) 0
Status check failed (system) 0

format that can be used through SIH, where there are destination. Te robot, depending on the available in-
several options shown in fgure (b) to deal with the image formation on the map, chooses the most appropriate route
and clarify it, where this process is a feedback on the nature for itself. Te robot can avoid collisions with fxed and
of the place and clarifcation of the image by the user and moving obstacles, where the distance depends on the lidar
where it is possible to save the image after modifcation in settings and ranges from 0.12 meters to 3.5 meters. Figure 14
S3 as well. shows the process of moving the robot from the starting
After downloading the map, it is now possible to direct point to the point to which it is intended and the path. Te
the robot from one place to another by suggesting the tuning guide for navigation is shown in Table 4.
Journal of Robotics 11

Table 4: Tuning guide for navigation in Figure 14.

Item Value
obstacle_range 3
raytrace_range 3.5
max_vel_x 0.22
min_vel_x - 0.22
max_vel_trans 0.22
min_vel_trans 0.11
max_vel_theta 2.75
min_vel_theta 1.37
acc_lim_x 2.5
acc_lim_theta 3.2
xy_goal_tolerance 0.05
yaw_goal_tolerance 0.17
sim_time 1.5
vx_samples 20
vy_samples 0
vth_samples 40

Turtlebot3 burger Remote PC

ROS_MASTER_URI = http://IP_OF_REMOTE_PC:11311 ROS_MASTER_URI = http://IP_OF_REMOTE_PC:11311

ROS_HOSTNAME = IP_OF_TURTLEBOT3 ROS_HOSTNAME = IP_OF_REMOTE_PC

Figure 15: ROS master running on a remote PC.

LDS-01 LDS-02
Figure 16: LDS-01 and LDS-02.
12 Journal of Robotics

Figure 17: Estimate initial pose.

Table 5: Specifcation of Turtlebot3 Burger.

Items Burger
maxUrange Te lidar sensor’s maximum practical range is set by this setting
Setting a value too large will necessitate more processing power for the map
map_update_interval
calculation; the smaller the value, the more frequently the map is updated
Te minimum score value that determines whether the sensor’s scan data matching
minimumScore test is successful or unsuccessful is established by this option. Tis can lessen errors
in the robot’s predicted position over a large area if confgured properly
linearUpdate When the robot has moved more than this distance, the scanning process will start
Te robot will begin the scan process if it detects more than this amount. Setting this
angularUpdate
number lower than linearUpdate is advised

Table 6: Tuning guide of Turtlebot3 Burger.

Items Burger
Te obstacle’s infation area is reduced by this parameter. Te route would be
infation_radius designed so that it would not pass through this region. Setting this to a value greater
than the robot’s radius is secure
Te cost value is multiplied by this factor. Because of the reciprocal relationship, the
cost_scaling_factor
cost decreases as this parameter is increased
max_vel_x Te maximum value of translational velocity is set for this factor
Te translational velocity’s minimum value is set for this factor. If this is set to zero,
min_vel_x
the robot can go backward
Maximum translational velocity value in reality; this is the fastest the robot is
max_trans_vel
capable of going
min_trans_vel Te actual translational minimum velocity; this is the fastest the robot can go
Maximum rotational velocity value in reality; this is the fastest the robot is capable of
max_rot_vel
going
min_rot_vel Te precise rotational velocity; this is the fastest the robot can go
acc_lim_x Te translational acceleration limit’s actual value
acc_lim_theta Te rotational acceleration limit’s actual value
xy_goal_tolerance Te permitted x, y distance when the robot is in its desired pose
yaw_goal_tolerance Te yaw angle is permitted once the robot has assumed its fnal position
Tis factor is advanced in the simulation by a few seconds. Too high a value prevents
sim_time quick rotation, whereas too low a value allows passage through a limited space.
Diferences in the length of the yellow line can be seen in the image below

5. Conclusion processing in the AWS cloud. Te practical results were

shown using a real TurtleBot3 Burger robot, the success of
Te approach proposed in this work is to solve the SLAM the process of creating a complex map containing 12 ob-
problem, controlling the robot via the AWS cloud, creating stacles of diferent sizes, where the robot was remotely
a map database, and storing it without the need to redraw the controlled by the EC2 server with certain specifcations, and
map, as well as using an authentication process (username after one hour, the EC2 monitor value was as follows: CPU
and password) under the supervision of the cloud admin- utilization: 26 percent, 8 MB network in, 9.26 MB network
istrator, and using the SIH to initiate cost-efective image out, network packets. In the 12.5 K count, network packets
Journal of Robotics 13

out: 12.2 k, status check failed 0. After that, the map has been ros-kinetic-move-base ros-kinetic-urdf ros-kinetic-
successfully stored in S3 for future use. Also, the process of xacro \
moving the robot from start to target inside the map was ros-kinetic-compressed-image-transport ros-ki-
tested, and the robot was able to move and avoid collision netic-rqt∗ \
with the obstacles it encounters on the way. From the
ros-kinetic-gmapping ros-kinetic-navigation ros-
foregoing, a person in a certain country can control a robot
kinetic-interactive-markers
located in another country, as this method ofers many
prospects for use in the military, medical, and educational Step 4: Set up the TurtleBot3 packages.
felds. For future work, the extended Kalman flter (EKF) $ sudo apt-get install ros-kinetic-dynamixel-sdk
algorithm can be used to improve the SLAM, and EKF-- $ sudo apt-get install ros-kinetic-turtlebot3-msgs
SLAM is used for various unknown environments. Ten,
intelligent optimization algorithms are suggested to improve $ sudo apt-get install ros-kinetic-turtlebot3
the performance of the mobile robot’s EKF-SLAM path. Step 5: Name your TurtleBot3 model.
$ echo “export TURTLEBOT3_MODEL � burger”
>> ∼/.bashrc
Appendix
Step 6: Network Confguration.
A. Detailed Steps for ROS in SLAM Figure 15 shows an example of network confgu-
and Navigation ration when ROS Master is running on
a remote PC.
(1) Installing ROS (PC setup)
Use the command below to connect your PC to
Te instructions were tested on Linux with Ubuntu a WiFi device and discover the assigned IP address.
16.04 and ROS Kinetic Kame, which are talked about
$ ifconfg
as follows.
Now, use the command listed in the following to
Step 1: Install ROS on remote PC
update the ROS IP settings after opening the fle.
First, the robotic operating system (ROS) is installed
$ nano ∼/.bashrc
on an Ubuntu Linux OS. Te steps to complete the
installation are clearly shown on the website https:// Place the cursor at the end of the line, then use the
www.ubuntu.com/. ifconfg command to replace localhost’s IP address in
the ROS MASTER URI and ROS HOSTNAME
In this work, the version of Ubuntu 16.04 LTS is used.
variables.
“Te Kinetic” is the codename for the latest ROS
distribution, which is compatible with Ubuntu 16.04. Use the command below to source the bashrc fle.
ROS is put on Ubuntu by using terminal software. To $ source ∼/.bashrc
start the installation, diferent steps were taken. (2) Single Board Computers (SBC) Setup
Step 2: Open the terminal and type each of the Step 1: TurtleBot 3 SBC Image Download For your
following instructions individually. hardware and ROS version, download the appro-
$ sudo apt-get update priate image fle. images using the Kinetic version of
$ sudo apt-get upgrade the Raspberry pi OS (raspbian OS).
$ wget https://raw.githubusercontent.com/ROBO Step 2: Te downloaded picture fle must be
TIS-GIT/robotis_tools/master/install_ros_kinetic.sh unzipped. Save the .img fle on the local disk after
being extracted.
$ chmod 755./install_ros_kinetic.sh
Step 3: Burn the image fle. Tere are many image-
$ bash./install_ros_kinetic.sh
burning tools available (the Linux disks utility and
Step 3: Install the required packages for ROS. raspberry pi imager are two examples).
$ sudo apt-get install ros-kinetic-joy ros-kinetic- Step 4: Te raspberry pi should start up.
teleop-twist-joy \
(a) Connect the HDMI cable from the monitor to
ros-kinetic-teleop-twist-keyboard ros-kinetic-
the raspberry pi’s HDMI port
laser-proc \
(b) Attach input devices to the USB port on the
ros-kinetic-rgbd-launch ros-kinetic-depthimage-to- raspberry pi
laserscan \ (c) Insert your microSD card
ros-kinetic-rosserial-arduino ros-kinetic-rosserial- (d) To start the raspberry pi, connect the power
python \ (either using USB or OpenCR)
ros-kinetic-rosserial-server ros-kinetic-rosserial-client Step 5: Confgure the Raspberry Pi
ros-kinetic-rosserial-msgs ros-kinetic-amcl ros-ki-
(a) Connect to the Wi-Fi network linked to the PC
netic-map-server \
once raspbian OS is up and running
14 Journal of Robotics

(b) Use the following command to determine the Step 1: Utilize the micro-USB connection to join the
raspberry pi’s assigned ip address. Te wlan0 OpenCR and the raspberry pi
section is typically where you will fnd the Step 2: To upload the OpenCR frmware, install the
raspberry pi’s wireless IP address necessary packages on the raspberry pi
$ ifconfg
$ sudo dpkg--add-architecture armhf
(c) Open the terminal on your PC, enter the ip
address of the raspberry pi, and connect. tur- $ sudo apt-get update
tlebot has been set as the default password $ sudo apt-get install libc6:armhf
$ ssh pi@{IP_ADDRESS_OF_RASPBERRY_PI} Step 3: Choose between the names Burger and Wafe
(d) After logging in, run the following commands on for the OPENCR MODEL depending on the
the raspberry pi to get the time in sync platform
$ sudo apt-get install ntpdate
$ export OPENCR_PORT � /dev/ttyACM0
$ sudo ntpdate ntp.ubuntu.com
(e) Open the confguration interface for the $ export OPENCR_MODEL � burger
raspberry pi $ rm-rf./opencr_update.tar.bz2
$ sudo raspi-confg Step 4: After downloading the loader and frmware,
(f ) Expand the fle system by choosing advanced extract the fle.
options, then click Exit.
(g) ROS network confguration $ wget https://github.com/ROBOTIS-GIT/OpenCR-
$ nano ∼/.bashrc Binaries/raw/master/turtlebot3/ROS1/latest/
(h) Modify the ip addresses for ROS MASTER URI opencr_update.tar.bz2
and ROS HOSTNAME at the fle’s end $ tar-xvf opencr_update.tar.bz2
export ROS_MASTER_URI � http://{IP_AD- Step 5: Uploading frmware to OpenCR
DRESS_OF_REMOTE_PC}:11311 $ cd./opencr_update
export ROS_HOSTNAME �
{IP_ADDRESS_OF_RASPBERRY_PI_3} $./update.sh $OPENCR_PORT
(i) After saving the document, close the nano editor $OPENCR_MODEL.opencr
(j) Apply modifcations using the following (4) Bringup
command (1) Run Roscore on a computer
$ source ∼/.bashrc $ roscore
Step 6: Confguration new lds-02 (2) Bringup TurtleBot3
Te TurtleBot3 LDS has been replaced by the LDS-02 Step 1: On a computer, launch a new terminal and
in models made in 2022. To use TurtleBot3’s SBC, use the raspberry pi’s ip address to connect.
adhere to the following directions. Te two cate- $ ssh pi@{IP_ADDRESS_OF_RASPBERRY_PI}
gories of leaders are shown in Figure 16.
Step 2: Bring up basic packages to start TurtleBot3
(a) Install the TurtleBot3 package update and the applications
LDS-02 driver $ roslaunch turtlebot3_bringup
$ sudo apt update turtlebot3_robot.launch
$ sudo apt install libudev-dev
$ cd ∼/catkin_ws/src (5) Teleoperation
$ git clone -b develop https://github.com/ Diferent remote controllers can be used to teleo-
ROBOTIS-GIT/ld08_driver.git perate the TurtleBot3 (Keyboard, RC100, PS3 Joy-
$ cd ∼/catkin_ws/src/turtlebot3 && git pull stick, XBOX 360 Joystick, Wii Remote). Verify that
$ rm-r turtlebot3_description/turtlebot3_teleop/ SBC and ROS versions support the required ROS
turtlebot3_navigation/turtlebot3_slam/ packages. In this study, keyboard teleoperation is
turtlebot3_example/ used. For keyboard teleoperation, start the Tur-
$ cd ∼/catkin_ws && catkin_make tleBot3 teleop key node from the remote PC. You
(b) Te LDS MODEL should be exported to the should substitute the name of your model, such as
bashrc fle. Use LDS-01 or LDS-02 depending on burger, wafe, or wafe pi, for the $TB3 MODEL
the model of LDS. parameter.
$ echo ‘export LDS_MODEL � LDS-01’ >> $ export TURTLEBOT3_MODEL �
∼/.bashrc ${TB3_MODEL}
(c) Use the command below to make the $ roslaunch turtlebot3_teleop
modifcations. turtlebot3_teleop_key.launch
$ source ∼/.bashrc
Te following set of instructions will appear in the
(3) OpenCR Setup terminal window if the node is successfully launched
Journal of Robotics 15

Control Your Turtlebot3mmc1 (7) Navigation

Moving around Te goal of navigation is to move the robot through
w a given environment from one location to the desired
location. A map with geometry data on the walls,
asd
furniture, and other objects in the environment is
x necessary for this. Te map was created using the
w/x: increase/decrease linear velocity sensor’s distance data and the robot’s own position
a/d: increase/decrease angular velocity data, as we discussed in the SLAM section earlier. A
robot can navigate from its current position to
space key, s: force stop a destination pose on a map by using the map, en-
CTRL-C to quit coder, IMU sensor, and distance sensor of the robot.
(6) SLAM Te following are the steps to complete this task.
Step 1: Launch the SLAM node from a new terminal Step 1: Launch the navigation
that is open in remote PC. Gmapping is the standard $ export TURTLEBOT3_MODEL � burger
SLAM technique. Frontier exploration, Hector, Karto, $ roslaunch turtlebot3_navigation tur-
and Cartographer are additional approaches. For the tlebot3_navigation.launch map_fle: � $HOME/
TurtleBot3 MODEL parameter, select the appropriate map.yaml
keyword from burger, wafe, and wafe pi.
Step 2: Estimate initial pose
$ export TURTLEBOT3_MODEL � burger
Prior to starting the navigation, initial pose esti-
$ roslaunch turtlebot3_slam turtlebot3_slam.launch mation is necessary since it initializes the crucial
Once the SLAM node is operational, TurtleBot3 will AMCL parameters. TurtleBot3 must be in the right
use teleoperation to explore uncharted territory on place on the map because the LDS sensor data neatly
the map. It is crucial to refrain from jerky motions, overlaps the map, as shown in Figure 17:
including abrupt changes in linear and angular
speed. It is a good idea to scan the entire map while (1) Click the 2D Pose Estimate button in the
using the TurtleBot3 to generate a map. Rviz menu.
(2) Drag the green arrow in the direction you want
Step 2: Tuning Guide for slam: Gmapping ofers the robot to face by clicking on the map and
a variety of parameters to adjust performance for setting the robot’s destination. A marking that
various settings. When confguring GMapping pa- can indicate the robot’s destination is this green
rameters, this tuning guide ofers advice. Te fle arrow. Te arrow is based on the x and y co-
TurtleBot3 slam/confg/gmapping params.yaml is ordinates of the target, while the angle is based
defned in Table 5 with the following parameters: on the direction of the arrow. As soon as x, y, and
Step 3: Save Map θ are set, TurtleBot3 will start moving to the
Te map is produced using the robot’s odometry, tf, destination immediately.
and scan data. Te map information was shown in (8) Te Tuning Guide for Navigation (shown in Table 6):
the Rviz window as the TurtleBot3 was moving. Save Te navigation stack has many parameters to change
the map data to the local drive after creating performance for diferent robots. Tis tuning is
a complete map of the area so you can use it later. defned in
Run the map saver node of the map server package to
turtlebot3_navigation/param/
generate map fles. In the directory where the map
dwa_local_planner_params_${TB3_MODEL}.yaml
saver node is launched, the map fle is saved. If no
other fle name is specifed, the word “Map” will be
used as the default and will create the fles map.pgm B. ROS for Obstacle Detection and
and map.yaml. Te -f option specifes a folder lo- Point Operation
cation and fle name when saving fles. Tat com-
mand will save Map.pgm and Map.yaml to the user’s Robot path planning (RPP) can be divided into two categories
home directory. based on the environment where the robot is located: static
(environment with fxed obstacles) and dynamic (the envi-
$ rosrun map_server map_saver-f ∼/map ronment has moving obstacles). Tese two categories can each
Step 4: Te two-dimensional occupancy grid map be further broken down into subgroups, including global path
(OGM), which is popular in ROS, is utilized for the planning (GPP), which allows for the advance knowledge of
map. Te saved map will resemble the following all fxed and moving obstacles. As a result, the GPP can be
illustration, with the white areas denoting colli- prepared in advance of the robot moving (ofine), and local
sion-free zones, the black areas denoting occupied path planning (LPP) is available. It is impossible to know
and inaccessible zones, and the gray areas denoting beforehand what the environment is like here. Terefore, as
uncharted territory. Te navigation is done using the mobile robot travels through the world, sensors (online)
this map. gather data about what is nearby [33].
16 Journal of Robotics

LDS data can move or stop the TurtleBot3. Te Tur- approach,” International Journal of Social Robotics, vol. 8,
tleBot3 burger stops moving when it sees something in its no. 3, pp. 371–380, 2016.
path. [remote PC] Start the fle with the obstacle. [11] R. K. Megalingam, C. R. Teja, S. Sreekanth, and A. Raj, “Ros
based autonomous indoor navigation simulation using slam
$ roslaunch turtlebot3_example algorithm,” 2019, http://www.ijpam.eu.
turtlebot3_obstacle.launch [12] G. Jiang, L. Yin, S. Jin, C. Tian, X. Ma, and Y. Ou, “A si-
For point operation, the TurtleBot3 can be moved using multaneous localization and mapping (SLAM) framework for
2D (x, y) points and z-angular movements. For in- 2.5D map building based on low-cost LiDAR and vision
fusion,” Applied Sciences, vol. 9, no. 10, 2019.
stance, TurtleBot3 moves to the point (x � 0.7 m,
[13] P. Arena, C. Bonomo, L. Fortuna, M. Frasca, and S. Graziani,
y � 0.2 m) and then rotates 60° if you enter (0.7, 0.2, 60). “Design and control of an IPMC wormlike robot,” IEEE
$ roslaunch turtlebot3_example Transactions on Systems, Man and Cybernetics, Part B (Cy-
turtlebot3_pointop_key.launch bernetics), vol. 36, no. 5, pp. 1044–1052, 2006.
[14] A. Buscarino, L. Fortuna, M. Frasca, and A. Rizzo, “Dynamical
network interactions in distributed control of robots,” Chaos:
Data Availability An Interdisciplinary Journal of Nonlinear Science, vol. 16,
no. 1, 2006.
Data sharing is not applicable to this article as no datasets
[15] D. J. Watts and S. H. Strogatz, “Collective dynamics of ‘small-
were generated or analyzed during the current study. world’ networks,” Nature, vol. 393, no. 6684, pp. 440–442,
1998.
Conflicts of Interest [16] S. Kajita and B. Espiau, “Legged robots,” in Springer Hand-
book Of Robotics, pp. 361–389, Springer Berlin Heidelberg,
Te authors declare that they have no conficts of interest. Berlin, Germany, 2008.
[17] U. U. S. Rajapaksha, C. Jayawardena, and B. A. MacDonald,
“Design, implementation, and performance evaluation of
References a web-based multiple robot control system,” Journal of Ro-
[1] F. H. Ajeil, I. K. Ibraheem, M. A. Sahib, and A. J. Humaidi, botics, vol. 2022, Article ID 9289625, pp. 1–24, 2022.
“Multi-objective path planning of an autonomous mobile [18] A. T. Rashid, M. Frasca, A. A. Ali, A. Rizzo, and L. Fortuna,
robot using hybrid PSO-MFB optimization algorithm,” Ap- “Multi-robot localization and orientation estimation using
plied Soft Computing, vol. 89, Article ID 106076, 2020. robotic cluster matching algorithm,” Robotics and Autono-
[2] A. Hussnain, Application of Cloud Robotics for Automatic mous Systems, vol. 63, pp. 108–121, 2015.
Manipulation, University of Tampere, Tampere, Finland, [19] J. S. Pershina, S. Y. Kazdorf, and A. V. Lopota, “Methods of
2018. mobile robot visual navigation and environment mapping,”
[3] M. Lacity and E. Whitley, “service automation: robots and the Optoelectronics, Instrumentation and Data Processing, vol. 55,
future of work hashtag for twitter users: #lserobots department no. 2, pp. 181–188, 2019.
of management public lecture,” 2016, https://www.lse.ac.uk/ [20] T. C. Smith, An Investigation on a Mobile Robot in a ROS
Events/2016/05/20160509t1830vOT/Service-Automation. Enabled Cloud Robotics Environment, Georgia Southern
[4] X. Ma and Y. Huang, “Research on mobile cloud robotics University, Statesboro, GA, USA, 2016.
based on cloud computing,” in Proceedings of the 2016 In- [21] G. Tuna, K. Gulez, V. Cagri Gungor, and T. Veli Mumcu,
ternational Forum on Management Education and In- “Evaluations of diferent simultaneous localization and
formation Technology Application, Guangzhou, China, mapping (SLAM) algorithms,” in Proceedings of the IECON
January 2016. 2012 - 38th Annual Conference on IEEE Industrial Electronics
[5] R. Siegwart, I. R. Nourbakhsh, and D. Scaramuzza, In- Society, pp. 2693–2698, Montreal, Quebec, October 2012.
troduction to Autonomous Mobile Robots, Massachusetts [22] J. Ahmed Abdulsaheb and D. Jasim Kadhim, “Robot path
Institute of Technology, Cambridge, MA USA, 2011. planning in unknown environments with multi-objectives
[6] J. Wan, S. Tang, H. Yan, D. Li, S. Wang, and A. V. Vasilakos, using an improved COOT optimization algorithm,” In-
Cloud Robotics: Current Status and Open Issues, IEEE Access, ternational Journal of Intelligent Engineering and Systems,
Piscataway, NJ, USA, 2016. vol. 15, no. 5, pp. 548–565, 2022.
[7] I. Afanasyev, A. Sagitov, and E. Magid, “Ros-based slam for [23] R. Amsters and P. Slaets, Turtlebot 3 as a Robotics Education
a gazebo-simulated mobile robot in image-based 3d model of Platform, Purdue University, West Lafayette, IN, USA, 2020.
indoor environment,” in Proceedings of the International [24] H. Deng, J. Xiong, and Z. Xia, “Mobile manipulation task
Conference on Advanced Concepts for Intelligent Vision Sys- simulation using ROS with MoveIt,” in Proceedings of the 2017
tems, pp. 273–283, Catania, Italy, October 2015. IEEE International Conference on Real-Time Computing and
[8] R. Doriya, P. Sao, V. Payal, V. Anand, and P. Chakraborty, “A Robotics (RCAR), pp. 612–616, Okinawa, Japan, July 2017.
review on cloud robotics based frameworks to solve simul- [25] S. Haider Abdulredah and D. Jasim Kadhim, “Developing
taneous localization and mapping (slam) problem,” 2015, a real time navigation for the mobile robots at unknown
https://arxiv.org/abs/1701.08444. environments,” Indonesian Journal of Electrical Engineering
[9] K. Takaya, T. Asai, V. Kroumov, and F. Smarandache, and Computer Science, vol. 20, no. 1, p. 500, 2020.
“Simulation environment for mobile robots testing using ROS [26] M. Robert and L. Avanzato, “Work in progress: introductory
and Gazebo,” in Proceedings of the 2016 20th International mobile robotics and computer vision lab-oratories using ros
Conference on System Teory, Control and Computing and matlab introductory mobile robotics and computer vision
(ICSTCC), pp. 96–101, Sinaia, Romania, October 2016. laboratories using ros and matlab,” in Proceedings of the 2018
[10] R. Limosani, A. Manzi, L. Fiorini, F. Cavallo, and P. Dario, ASEE Annal Conference and Exposition, Salt Lake City, UT,
“Enabling global robot navigation based on a cloud robotics USA, June 2018.
Journal of Robotics 17

[27] F. Penné, Implementation of Particle Filtering in a Turtlebot,

University of Chicago, Chicago, IL, USA, 2015.
[28] H. Wang, W. Liu, F. Zhang, S. X. Yang, and L. Zhang, “A GA-
fuzzy logic based extended Kalman flter for mobile robot
localization,” in Proceedings of the 2015 12th International
Conference on Fuzzy Systems and Knowledge Discovery
(FSKD), pp. 319–323, Zhangjiajie, China, August 2015.
[29] J. Nadon, “Your content solution: an introduction to AWS
S3,” in Website Hosting and Migration with Amazon Web
Services, pp. 15–24, Apress, Berkeley, CA, USA, 2017.
[30] R. Saini and R. Behl, “An introduction to aws—ec2 (elastic
compute cloud),” 2020, https://aws.amazon.com/pm/ec2-amd/?
trk=17a47518-9bfe-4f1e-a67f-42bdd19264e5&sc_channel=ps&
ef_id=EAIaIQobChMI3-fCvL3s_QIV2yMrCh20AArOEAAYAS
AAEgKkCvD_BwE:G:s&s_kwcid=AL!4422!3!525855180410!p!!g
!!amazon%20web%20server!13385427590!122597168825.
[31] Amazon, “What is amazon EC2? - amazon elastic compute
cloud,” 2023, https://docs.aws.amazon.com/AWSEC2/latest/
WindowsGuide/concepts.html.
[32] Amazon, “Fast and cost-efective image manipulation with
serverless image handler | aws architecture blog,” 2023,
https://aws.amazon.com/blogs/architecture/fast-and-cost-
efective-image-manipulation-with-serverless-image-
handler/.
[33] H. Miao and Y.-C. Tian, “Robot path planning in dynamic
environments using a simulated annealing based approach,”
in Proceedings of the 2008 10th International Conference on
Control, Automation, Robotics and Vision, pp. 1253–1258,
Hanoi, Vietnam, December 2008.