Internet of Robotic Things: Current Technologies,

Applications, Challenges and Future Directions

Davide Villa, Xinchao Song, Matthew Heim, Liangshe Li
Department of Electrical and Computer Engineering
Northeastern University
Boston, MA, 02115 USA
{villa.d, song.xin, heim.m, li.liangs}@northeastern.edu

Abstract—Nowadays, the Internet of Things (IoT) concept repetitive and tedious jobs in a manufacturing line to helping
arXiv:2101.06256v1 [cs.RO] 15 Jan 2021

is gaining more and more notoriety bringing the number of and performing critical or dangerous tasks unaffordable by
connected devices to reach the order of billion units. Its smart a human being, such as rescuing in disaster areas or extra-
technology is influencing the research and developments of
advanced solutions in many areas. This paper focuses on the terrestrial operations. The robots initially used in these appli-
merger between the IoT and robotics named the Internet of cations were just single machines with limitation given by their
Robotic Things (IoRT). Allowing robotic systems to communicate hardware components and computational abilities. To resolve
over the internet at a minimal cost is an important technological those issues, the robots were started to be connected in a
opportunity. Robots can use the cloud to improve the overall communication network through a wired or wireless creating a
performance and for offloading demanding tasks. Since commu-
nicating to the cloud results in latency, data loss, and energy loss, Networked Robotic System [7]. However, they suffered from
finding efficient techniques is a concern that can be addressed inherent resource constraints which led to network latency,
with current machine learning methodologies. Moreover, the limited memory and low computational and learning capa-
use of robotic generates ethical and regulation questions that bilities. Today real-life scenarios demand fast and complex
should be answered for a proper coexistence between humans task executions which require sophisticated data analysis and
and robots. This paper aims at providing a better understanding
of the new concept of IoRT with its benefits and limitations, as high computational abilities. The solution to these limitations
well as guidelines and directions for future research and studies. have been recently tried to be addressed in the novel form
of Cloud Robotics [8] which takes advantage of a cloud
Index Terms—Internet of Things (IoT), autonomous systems, infrastructure to access resources on-demand and to support
network communication, Internet of Robotic Things (IoRT), the operations. However, these systems are affected by new
regulations.
issues like interoperability, network latency, security, quality
of service and standardization [9]. The Internet of Robotic
I. I NTRODUCTION
Things perfectly fits in this scenario aiming to overcome these
Over the past decades till today, the Internet of Things is constraints by combining together the IoT with robotics and
experiencing an exponential growth and attention [1] [2]. Its leading to a more efficient, smart, adaptive and also cheaper
smart technology allows the creation of a network of real- robotic network solution.
world objects, namely ‘Things’, with the purpose of connect- This paper has the aim to provide a comprehensive de-
ing everything through the ability of exchanging information scription of the Internet of Robotic Things concept showing
over the Internet [3] [4]. This revolution is shaping all aspects its general architecture, benefits and challenges as well as
of the human life. The IoT concept is being adopted by more some practical use cases and guidelines in the hope of giving
and more organizations in several fields, e.g. military, robotics, awareness on this subject and inspiring future research. The
healthcare, nanotechnology or space, creating the Internet of rest of the paper is organized as follows. Section II gives an
X Things, where X is the relevant area. Currently, according overview of the IoRT concept, while Section III explains in
to the analyst firm Gartner [5], 127 new IoT devices are more details its architecture. Some practical use-case examples
connected to the Web every second bringing its overall number are discussed in Section IV. Section V addresses the main
to reach 75 billion units by 2025. The global spending on the challenges that the current research is facing during the devel-
IoT should reach 1.29 trillion dollars during 2020 with a 93% opment of an IoRT system. Moreover, Section VI introduces
of adoption of IoT technology by the enterprises. the ethical issues that arise in the use of IoRT in the everyday
An area where the Internet of Things technology is finding life, while Section VII and Section VIII discuss the topics
fertile ground is the robotics. Robotics consists of a mod- of regulations and how IoRT can shape the modern society.
ern and fast-evolving technology that is bringing enormous Finally, the conclusions are addressed in Section IX.
changes in several aspects of human society over the past
decades. It can be defined as ”The branch of engineering that II. OVERVIEW OF THE I NTERNET OF ROBOTIC T HINGS
involves the conception, design, manufacture and operation of The Internet of Robotic Things (IoRT) is the merge between
robots” [6]. Its application can range from the execution of IoT and robotics. The term IoRT was coined by Dan Kara in a
report of ABI Research [10] to denote intelligent devices that
can monitor events, gather data from a variety of sources and Service and Application
sensors, and exploit both a local and a distributed intelligence • Application protocols
• Cloud computing
to control objects and determine the best action to follow. More • Machine learning
in general, the IoRT can be seen as a global infrastructure en-
abling advanced robotic services thanks to the interconnection
of robotic things. The robotic things can exploit the benefits of
modern communication and interoperable technologies based Network and Control
on the cloud, through the IP protocol and its IPv6 version, to • Communication protocols
take advantages regarding data processing, memory storage, • Controllers
• Cloud data storage
computational overhead and security. This mitigates and can
resolve the previous robotics issues. Moreover, the Internet of
Robotic Things goes beyond the networked and cloud robotic
paradigms by taking advantage of the enabling IoT tech- Physical
nologies. This aims at integrating heterogeneous intelligent • Robots
devices empowering enormous flexibility in the design and • Sensors
implementation of a distributed architecture which provides • Actuators
computing resources both in the cloud and at the edge.
The IoRT multidisciplinary nature follows the same trend of
the IoT by providing advanced robotic capabilities leading to Fig. 1. Internet of Robotic Things architecture overview.
the rise of interdisciplinary solutions for various and different
disciplines. From a technological perspective, IoRT enhances sensors refer to any kind of system that the agent can use
the ordinary robot abilities which are usually classified in 3 to perceive a data from the surrounding environment, while
groups: basic (perception, motion, manipulation); higher-level actuators to any kind of tool used to perform an action. Both
(decisional, autonomy, interaction cognitive); and system-level sensors and actuators can also be integrated inside a single
(configurability, adaptability, dependability). To achieve this, robot system to enhance and optimize the performance. As in
in addition to the IoT features, IoRT solutions are supported the standard OSI model, the purpose of the physical layer is
by other technologies such as: multi-radio access to link smart to abstract all of technical information and provide a set of
devices together; artificial intelligence to generate optimized services about its periphery to the network and control layer.
solutions for complex problems; and cognitive technologies to
allow operational efficiency. B. Network and Control
To summarize, the Internet of Robotic Things is designed to The Network and Control layer provides an efficient in-
be placed at the summit of the Cloud robotics and combines tegration of the hardware components of the physical layer
the IoT technology features with the autonomous and self- by means of control processes and communication protocols.
learning behavior of connected robotic things in order to Robotic devices are able to send information to other robots
generate advanced and smart solutions making an optimal use or to the cloud using various methods. There are a few main
of distributed resources. protocols used for communication depending on the range
between sender and receiver and the power required to send
III. I O RT ARCHITECTURE data over that range. Using any protocol should allow data to
The architecture of an Internet of Robotic Things system be communicated over the Internet. At the link and physical
can be described as being composed by three main layers: layers, common short-range protocols are IEEE 802.11 (Wi-
Physical, Network and Control, Service and Application (Fig- Fi) and Bluetooth. Common long-range protocols are Sigfox,
ure 1) [11] [12]. Its composition recalls the typical structure LoRaWAN, and NB-IoT. Another protocol worth noting is
of the OSI model but with a different perspective which takes IEEE 802.15.4 which makes the basis for 6LowPAN and
into account also the robotic part of IoRT. Zigbee. At the network layer, common protocols used for IoT
are 6LowPAN and the Routing Protocol for Low Power and
A. Physical Lossy Networks (RPL). Table I shows applicable protocols for
The physical layer consists of the lowest part of the IoRT robotics at different layers of the protocol stack.
architecture. It includes everything related to the hardware 6LoWPAN allows a Low Rate Wireless Personal Area
components, such as robots, sensors and actuators. The robots Network (LR-WPAN) to use IPv6. This protocol is particularly
can be described as intelligent agents that can communicate used in the IoT for its ability to be low-energy and low-cost
between each other and have the ability to establish a more which is crucial for small, battery-powered robots or sensors.
complex system in order to accomplish a specific objective Protocols that were initially designed for wired networks,
thanks to a sequence of distributed actions. A robot can be such as Precision Time Protocol (PTP), can be adapted for
represented for example with a vehicle, a drone, a home 6LoWPAN [13]. RPL, also using IPv6, creates a directed
appliances, or a health care equipment. On the other hand, acyclic graph to structure the network and allows three types

2
 TABLE I may be subjective trustworthiness, where every node in a
I O RT PROTOCOL STACK network computes their own trust of other nodes. In contrast,
Layer IoRT protocols Usage
there may be objective trustworthiness, where the trustworthi-
ness of a node lies in a distributed system that is shared among
Application CoAP, MQTT, REST API, Data streaming
XMPP, TLS the nodes in the network. Subjective trustworthiness can slow
Transport UDP, TCP End-to-end communi- a network down but can be more protected from malicious
cation intent than an objective trustworthiness system.
Network IPv6 Internet connection Looking to the future, one area of research that shows
Link Wi-Fi, 802.15.4, LoRaWAN, Robot-to-Robot, promise in the IoRT is Large Intelligent Surfaces (LIS) [17].
NB-IoT, Bluetooth, Sigfox Access Point, gateway Walls can have a large number of antennas and act as a massive
communication multi-input multi-output (MIMO) system. This technology
Physical Robots, Sensors, Actuators Data acquisition could allow for more capabilities in a smart environment by
making the infrastructure more active. Robots could receive
of communication: Multi-Point-to-Point, Point-to-Multi-Point, data from large intelligent surfaces and could even interact
and Point-to-Point. with them in the case of a wall-climbing robot.
Sigfox, a low-power wide-area network (LPWAN), works
with Ultra Narrowband (UNB) signals at frequencies lower C. Service and Application
than 1 GHz with a very low datarate of up to 100 bits/sec [14]. The Service and Application layer resides at the top of
The result is a very low energy signal that can pass through the IoRT architecture and relies on program implementations
buildings and reach underground. This technology could be to properly perform the end-user operations of monitoring,
effective for robots to send short status messages from a long processing and controlling. There are several options for
distance, but they must talk through a mobile gateway first. protocols. One example is CoAP (Constrained Application
LoRaWAN is another LPWAN that uses chirp spread spec- Protocol) which uses UDP at the transport layer. It is useful for
trum (CSS) around the same frequencies of Sigfox. However, resource-constrained devices on low-power, lossy networks.
the two technologies do not interfere with one another since The security and applicability of using CoAP for IoT devices is
the power of LoRaWAN signals are spread over a band of questionable. A more widely used protocol is MQTT (Message
frequencies. LoRaWAN uses different spreading factors that Queue Telemetry Transport) which uses a publish/subscribe
affect the datarate and possible range of transmission. With architecture. MQTT aims to be lightweight and reliable over
spreading factors 7 to 12, each one is orthogonal to the others. unreliable networks. A commonly used communication archi-
While Sigfox and LoRaWAN works on unlicensed fre- tecture in the IoT is REST, or REpresentational State Transfer.
quency bands, NB-IoT works on a licensed spectrum which It only relies on simple HTTP methods (GET, POST, PUT)
guarantees more reliability with less interference. In an anal- and its ubiquity is a strength. However, REST adds to the
ysis of offshore unmanned aerial vehicle (UAV) monitoring complexity of a robotic system and may be avoided. Another
cargo vessels, NB-IoT base stations (BS) could be used common application protocol is XMPP (eXtensible Message
effectively on UAVs acting as relays [15]. Localization for Presence Protocol). It is an open standard that aims to send
UAVs is an ongoing research area addressing problems such close to real-time messages between devices in a decentralized
as network failure when losing line of sight with the UAV and manner. As a case study, a cloud platform for an exoskeleton
GPS limitations. robot chose to use MQTT to transmit data to the cloud and
Security can be addressed at all levels of the protocol stack: HTTP to communicate from the cloud [18].
application, transport, network, link, and physical. Covering An integral part of current robotics is machine learning
the application and transport layers, the Transport Layer Secu- techniques that can be used for tasks such as mapping, lo-
rity (TLS) protocol can provide secure and trusted data trans- calization (knowledge of a robot’s own location), and learning
mission by applying message integrity, authentication, and the environment. Machine learning has traditionally been used
confidentiality. At the network layer, RPL can be used in pre- for object detection, tracking, and classification. However,
installed and authenticated modes for message integrity and its applicability to robotic learning means it is continually
secure communication, respectively. At the link and physical adapting field being used to solve more complex problems.
layers, Wi-Fi can also use authentication and data encryption. Cloud robotics is used to keep robotic things operating at
The use of frequency hopping and spread spectrum can allevi- a low power and intelligence level while having a powerful
ate the problems of interference and jamming. However, these data center to communicate with. Robots may send data to the
examples only cover a small subset of the available protocols, cloud for mapping and localization, perception, and actuation
so security is an ongoing critical issue with IoT devices. purposes. The cloud can be accessed for more data that robots
Many of the application areas for the IoRT, such as man- may need for training neural networks. The robot can also
ufacturing, are high-stakes environments and therefore should learn by taking models from the cloud without having to per-
have cybersecurity measures to protect the network from form extra computations. The cloud may provide more general
malicious intent. One way to address the problem of malicious knowledge while the robot learns its local environment. Cloud
nodes in a network is trustworthiness management [16]. There robotics also allows for human-robot interaction by having

3
humans make decisions for robots or control robots over a 1. Sensory Data Collection
network. Having multiple robots communicate over a network
simultaneously can result in high latency and data loss. There Remote : Cloud Local : Workstation Field : Robot
may be much more data that robots need to send to the cloud
2. Model Training 4. Testing Embedded
or to other robots. For example, streaming LIDAR over Wi- 3. Model Validation
Deploy IoT Data IoT Data
Fi can result in a network failure with multiple robots [19].
Trained Recorder Recorder
Robots can use deep reinforcement learning (RL) to query the Model
cloud intelligently with a minimal frequency and associated
cost.
It is important for a robot that comes in contact with humans Fig. 2. IoRT in disaster response use case example.
to be able to act and react with humans differently than other
directly deployed in the robots so that they are immediately
robots. To accomplish this, a robot must express personality
able to successfully accomplish the rescue mission in that area.
and connect with the emotions and storytelling of humans
As an example, in the event of an earthquake, the robots
so that humans can understand what the robot is doing and
might be able to detect the victim’s position thanks to their
for what reason. A robot whose behavior is influenced by
sensors and to help on removing reinforced concrete that might
a robot can be called a socially responsive robot [20]. A
have collapsed. Additionally, If a victim is trapped under the
promising technique is when a human can teach a robot the
ruins and is not able to get out momentarily, the robots can
movements of a behavior by showing an example. RL can be
be responsible for transporting food and water to help the
used by the robot to learn the actions and states that make up a
victim maintain the nutrition they need. The overall success
behavior, such as dancing, without understanding the dynamics
rate of rescue operations and the survival rate of victims can
of systems. Effective application areas of socially responsive
be greatly improved thanks to the use of these robots.
robots include hospitals, rehabilitation, and generally smart
environments or indoor areas with many types of sensors. B. Precision agriculture
IV. A PPLICATION DOMAINS Authors in [22] provide an IoRT application in the field
IoRT has been widely used in many fields and the machine of precision agriculture. Figure 3 presents a sketch of an
learning concept is undoubtedly one of the key links. In agricultural IoRT system. First of all, the mobile robots in
particular, robots combined with IoT can bring lots of benefits the environment collect data using on-board sensors. The data
in doing repetitive tasks and in operating in disaster areas contain information about temperature, humidity, pressure and
where it is unaffordable for a human being to work properly. light measurements. The server can retrieve these data from
the mobile robots using cellular communication or Wi-Fi in
A. Disaster response order to process them. After that, the data will be deployed to
In [21] the authors exploit an IoRT architecture for a disaster web applications and the user is able to visualize them through
response. A lot of major disasters happen on Earth every year, a web interface.
e.g. earthquakes, tsunamis or typhoons. When a calamity of This solution brings the advantages of the IoRT into full
this magnitude happens, the time is very precious. The most play. The mobile robots are able to collect data on-demand
urgent task to achieve is to save as many people as possible instead of the human. These processed data can be used to
in the shortest time. Robots can offer a lot of help in this develop appropriate agricultural plans during climate changes
kind of situations after they have received a deep learning
training. In order to accomplish this, the first step is to use
mobile robots to collect data from the local environment. A Environment Server User
remote network elaborates these data to build an AI model
which is evaluated by the internal system. The model is also
distributed from the cloud to the local workstation for further
performance testing. After it has reached a certain level of
reliability, the model is ultimately deployed into the robots for Laptop
Agricultural Field Cloud Server
the next level of learning process. The proposed architecture
can be schematically represented in Figure 2.
This solution can make the robots have enough experience
on how to respond to disasters in one area. However, different
Robots inside the field Modem Local Server Mobile Devices
places might present different environments. If the robot has
information about the environment of only one single type of
place, it cannot be appointed to other places to carry out rescue Data Communication Back-End of Front-End of Browser or
Acquisition over Wi-Fi/Cell the Web App the Web App Application
missions. A solution to this issue is to consider to store in the
cloud different AI models for various environments. Once a
disaster happens, the AI model of that environment can be Fig. 3. Conceptual design of the agriculture system.

4
or hazard situations. This will result in a saving of time and In short, the IoRT applications are being largely considered
resources as well as an increasing of the productivity. in the academic field and some solutions have been already
However, static robots or devices are also needed in this used to develop a commercial product. With the imminent
system to measure weather conditions or provide alarm signals introduction of 5G and, in the future, of 6G, the IoRT
in case of hazards. An appropriate website design is a key technology has the potential to increase its performance and
point to present information handily and to visualize data in to continue to be considered a valuable field of research.
a more apprehensive way to the user. Further work can also
V. T ECHNOLOGY C HALLENGES
focus on the subsequent application of the processed data. For
example, the users could assign further instructions to mobile Although growing rapidly, IoRT is still a novel area in its
robots through web applications after the data acquired have preliminary stage and is facing many challenges that need to be
been collected and analyzed. overcome. This section introduces two of the most significant
challenges, i.e. data processing and security, and discusses
C. Medical services possible solutions to these problems.
Another field where IoRT can bring benefits is the medical A. Data Processing
one. In [23] authors have developed an application comprising Because of the dual cyber-physical nature of IoRT, huge
a massage robot. The robot exploits an IoT architecture with amounts of data are collected in various forms and from
sensors to collect data real-time and combine them with a different sources, e.g. data from various types of sensors
smart control system to apply the proper pressures of the or information for robot control. These data need to be
desired massage technique. The robot is a mixture of data- transmitted from the nodes to the cloud where they can be
mining and brain-based networks. This use case perfectly remotely processed. This requires a higher capability of data
represents the current trend of the technology showing how processing than a regular IoT application, and consequently
the IoRT is fitting in the medical health area. This opens the may cause the following challenges.
doors for a huge variety of future development research.
• Massive data streaming and processing usually demand
a significant communication bandwidth and a powerful
D. Smart Restaurant
computing capability in an IoRT network. As an example,
The revolution in the use of IoRT is shaping all aspects in a group of security robots may need to send different
the people’s daily life. In [24] a robot acting as a restaurant visual and acoustic data at the same time during their
waiter is proposed. Figure 4 shows the main blocks of the patrol. This requires fast data processing to make sure
system. The information about table number and menu are the system can rapidly respond to possible incidents.
stored in a QR code. The orders are managed by a website • Latency on real-time applications has a considerable
and received by both the cuisine and the manager together impact on the operating efficiency and safety. Since many
with the table number. As soon as the orders are ready, the IoRT systems, such as a robotic transpiration system in
chef utilizes the robot to deliver them to the customers. The a smart factory, are actively interacting with the environ-
whole system is automatic and the information are stored in ments in a time-sensitive manner, undesired latency could
the cloud. cause operation failure or even accidents.
The solution is simple and feasible. However, some key • Unstable communication in a complex environment may
problems still need to be addressed. For example, the robot significantly reduce the system performance. For exam-
lacks some emergency measure and it may have issues if the ple, a medical IoRT system in a hospital may suffer
walking route is blocked. Another significant concern is that from the hospital building’s complex indoor space and
the waiter robot is not able to speak with customers, which interference from other medical devices such as Magnetic
may leave a bad impression at the end. Nevertheless, this IoRT Resonance Imaging (MRI).
solution has the potential to significantly increase the employer To address those issues, possible solutions may include
profit by reducing the personnel expenses and to provide a using different computing frameworks, new communication
faster service to the customers. technologies, and optimization.
Edge computing and fog computing are two of the lat-
est popular solutions towards computational challenges. Both
Serve
Call Robot architectures aim at distributing the data computation and
Customer
storage all over the network, so that part of the data processing
Chef
QR Code Menu Order
can be done closer to the end nodes. The difference between
them is that edge computing is trying to process more data
locally, while fog computing is using IoT gateways to do that.
IoRT systems might exploit one or both technologies based on
Manager the working environment and requirement specifications [25].
As for communication, heterogeneous IoT technologies may
Fig. 4. Main blocks of the waiter robot system. be applied. It represents a multifaceted network integrating

5
different communication architectures, including Wi-Fi, cel- and backups, could help to guarantee the safety and robustness
lular networks, and Wireless Sensor Networks (WSN). It of the robotic part of the IoRT architecture.
can provide more choices and flexibility for massive data
exchanges among different types of nodes [4]. VI. E THICAL I SSUES
Another possible measure is to use higher-speed wireless As an integration of IoT and Robotics, IoRT is also facing
communication, such as Terahertz communication. Terahertz ethical issues and challenges from both sides. There are three
is the next generation of wireless communications which aims main ethical issues that should be taken into consideration:
to use terahertz band (0.3 THz - 10 THz) to unlock significantly robot ethic, property rights of information, and data privacy.
wider bandwidth and achieve higher communication speed It is crucial to address those issues properly for the harmonious
estimated to exceed 100 Gbit/s. It will be fast enough in coexistence between robots and humans in the modern society.
order to allow massive data exchange in most common IoRT
A. Robot Ethic
networks.
Finally, good optimization frameworks, which take pre- The ethic of a robot system is a topic that has been discussed
processing data, computation modes, and communication tech- since the dawn of the robot concept. The famous Three Laws
nologies together into consideration, can help to obtain the of Robotics by Isaac Asimov state [28]:
optimal computation strategies making the whole IoRT more 1) A robot may not injure a human being or, through
efficient [25]. inaction, allow a human being to come to harm.
2) A robot must obey orders given it by human beings except
B. Security and Safety
where such orders would conflict with the First Law.
Security is a crucial issue in this area. Many application 3) A robot must protect its own existence as long as such
domains of IoRT have high demands on safety and stability protection does not conflict with the First or Second Law.
among their operations. Security is also a challenging problem
The core idea here is clear: robots should benefit but not
since it is related not only to the security of the Internet of
harm users. However, as a technology with ethical impli-
Things but also to the safety of the robots. As we discussed
cations, IoRT cannot achieve this goal by itself but needs
above, there are massive amounts of various data in the IoRT
the relevant society, including robot manufacturers, users, and
systems that need to be exchanged between the robots and
governments, to take actions on effective policies and novel
the cloud as well as between robot and robot. This leaves
practices to prevent the misuse of robots.
more room for security failures, from leaking sensitive data
to cyber-attacks. Some typical problems are listed below [11] B. Property Rights
[26]. One of the ethical questions on data faced in the IoRT
• Insecure communication between cloud and robot or area regards the origin of these information and the owner
between robot and robot may cause sensitive data to of them. Since artificial intelligence and robots are involved
be exposed or hacked. The less the communication is and generate intelligent information, the answer could be even
secured, the easier a hacker breaks in, especially for those more tricky than for common IoT applications. As an example,
professional hackers sponsored by business competitors one of the difficulties is how to properly identify the authors
or even hostile foreign governments. of the data [29], which may not be clear in many cases.
• Authentication failure may allow hackers or unauthorized
personnel to easily obtain access to the system, hijacking C. Privacy
its parts or even the whole architecture, and sabotaging Another possible ethical question can be who should access
the robots by reprogramming them. the data and how these information should be used. Especially
• Robot failures, no matter whether caused by unexpected for sensitive data, such as medical or financial data, things
hardware faults or intentionally cyber-attacks, may dam- get more serious. According to the current society consensus
age the environment, injure people, or even cause sec- on privacy, no data collector should have unlimited and
ondary accidents. unregulated rights to access or use any data without proper
Those issues can be addressed and mitigated by the follow- permission or authorization. Due to the high capability of
ing measures. The first one is to use encrypted communication most IoRT applications to collect data from both their working
based on protocols like TLS and SSH to secure communi- environment and interacted clients, this issue needs to be
cation. New technologies like quantum communication may addressed carefully.
also be a future direction to provide a possible once and for
all solution to that. Additionally, strengthening authentication VII. R EGULATIONS
by using different measures such as data fragments, exter- From the above discussion, the necessity to have regulations
nalized authorization, or granular permission, can improve for the IoRT area can be clearly realized [30]. However, by
the authentication security [27]. Some new technologies like 2019, there was still no clear regulatory framework specifically
block-chain could also provide further assistance. Last but on IoRT, but only partial regulations that may have been
not least, safety measures on robots, such as hardware failure applied on relevant technologies [31]. As an example, in the
monitoring, emergency detection and stop, system redundancy European Union, robots can be regulated as general products

6
 Regulations

Autonomous
robots
Robot Ethic Property Rights Privacy Standards
Big data analysis Simulation

Uniform
IoRT
IoRT
Ethics Augmented Horizontal and vertical
Interface Industry 4.0
reality system integration

Fig. 5. The relationship between regulations and IoRT ethics and technologies.

under safety regulations Directive 2001/95/EC. Depending on

Additive The Industrial
their use, such as industrial robots or medical devices, they manufacuring Internet of Things
may also be subject to some specific legislation like Machinery
Directive 2006/42/EC or Regulation 2017/745. Regarding data
security, General Data Protection Regulation (GDPR) has The Cloud Cybersecurity
regulated how to properly process sensitive data. However,
since all those regulations are only for general robots or data Fig. 6. Nine key technology in Industry 4.0 [33].
alone, they cannot reflect and regulate the unique and specific
needs in the area of IoRT which combine robots and IoT
technologies together. In particular, due to the cyber-physical and robots are widely deployed and connected through IoT all
nature of IoRT systems, in many cases it is vague whether across the urban area. This involves an automation of public
they should be classified as products or services. Since both of facilities leading to more accessible and convenient services
them are usually regulated separately, it is sometimes hard to for all citizens.
determine the liability when an incident occurs. The negative
impacts are not only on ethical issues but also on technologies. IX. C ONCLUSION
The absence of regulatory frameworks leads to a lack of well-
defined standards or uniform IoRT technological interfaces, The Internet of Robotics Things is an emerging area aiming
which brings significant difficulties to integrate together dif- at integrating robotic technologies into IoT environments.
ferent IoT and robotic devices from various manufacturers, Internet of Robotic Things enables robotic systems to com-
especially when they are highly divisive and heterogeneous municate with each other, connect to the cloud, and share data
[12]. and knowledge to accomplish sophisticated tasks. In order to
The current situation could be improved by issuing new provide a better understanding of IoRT and its current develop-
mandatory regulations from governments or voluntary stan- ment directions, in this paper, we have reviewed its relevant
dards from industrial communities, although this might cre- concepts, technologies, applications, and current challenges,
ate other challenges regarding, for example, how to clarify as well as provided a future vision of what the IoRT can lead
liability [31]. As the usage of IoRT technologies becomes to. We have first discussed the three main layers that form
increasingly popular, this will be eventually solved through the IoRT architecture, i.e. physical, network and control, and
the collaboration of policymakers, companies, and consumers. service and applications. Then, we have presented four real
Future regulations should improve ethics and build a uniform use-case applications to show IoRT systems’ capabilities and
technological interface for IoRT as shown in Figure 5. what they can bring to modern society. Next, several critical
technological challenges, including data processing, security,
VIII. F UTURE V ISION and safety, have been discussed, and possible solutions have
The Internet of Robotic Things seems to have all qualities been proposed. We have also introduced some sensitive ethical
and prerequisites to become one of the core technologies to issues that need to be properly addressed to ensure a success-
support the future smart society. As an example, IoRT is the ful co-existence between robots and humans. Moreover, the
cornerstone for the imminent smart manufacturing [32] and current lack of regulations may negatively impact the further
stands as one of the nine key technologies that form the development of IoRT and, for this reason, this should be solved
foundation of Industry 4.0 [33] (shown in Figure 6). It may through the efforts of the whole society. Finally, as we have
even lead to the realization of the so called ”zero-labor factory” presented in the last part of the paper, we consider the Internet
which aims to fully automate manufacturing. of Robotic Things as a vital technology that has the potential
Likewise, the integration of the two technologies of Internet to bring enormous advantages in modern society and that can
of Things and robotics makes IoRT to perfectly fit in the contribute to the creation and development of the future smart
current trend of the smart city [34], where massive sensors manufacturing and smart cities.

7
 R EFERENCES [26] N. Dragoni, A. Giaretta, and M. Mazzara, “The internet of hackable
things,” in Proceedings of 5th International Conference in Software En-
[1] P. Simoens, M. Dragone, and A. Saffiotti, “The internet of robotic gineering for Defence Applications, pp. 129–140, Springer International
things: A review of concept, added value and applications,” International Publishing, 2018.
Journal of Advanced Robotic Systems, vol. 15, February 2018. [27] A. Wegner, J. Graham, and E. Ribble, Cybersecurity for Industry
[2] R. Batth, A. Nayyar, and A. Nagpal, “Internet of robotic things: Driving 4.0: Analysis for Design and Manufacturing, ch. A New Approach to
intelligent robotics of future - concept, architecture, applications and Cyberphysical Security in Industry 4.0, pp. 59–72. Springer International
technologies,” in 2018 4th International Conference on Computing Publishing, 2017.
Sciences (ICCS), pp. 151–160, August 2018. [28] S. L. Anderson, “Asimov’s’ three laws of robotics’ and machine
[3] S. Sabah, N. Qarabash, and H. Obaid, “The road to the internet of things: metaethics,” AI SOCIETY, vol. 22, no. 4, pp. 477–493, 2008.
a survey,” in 9th Annual Information Technology, Electromechanical [29] D. Popescul and M. Georgescu, “Internet of things–some ethical issues,”
Engineering and Microelectronics Conference (IEMECON), pp. 290– The USV Annals of Economics and Public Administration, vol. 13,
296, March 2019. pp. 210–216, December 2013.
[4] I. Ud Din et al., “The internet of things: A review of enabled technolo- [30] S. Loke, “Are we ready for the internet of robotic things in public
gies and future challenges,” IEEE Access, December 2018. spaces?,” pp. 891–900, Association for Computing Machinery, 10 2018.
[5] https://www.gartner.com/, November 2020. [31] E. Fosch Villaronga and C. Millard, “Cloud robotics law and regulation
[6] J. L. Fuller, Robotics. USA: Prentice Hall PTR, 2nd ed., 1998. cloud robotics law and regulation. challenges in the governance of
complex and dynamic cyber-physical ecosystems,” Robotics and Au-
[7] N. Mohamed, J. Al-Jaroodi, and I. Jawhar, “A review of middleware for
tonomous Systems, vol. 119, pp. 77–91, December 2018.
networked robots,” The International Journal of Computer Science and
[32] X. Liu, “Research toward iot and robotics in intelligent manufacturing:
Network Security, vol. 9, January 2009.
A survey,” International Journal of Materials, Mechanics and Manufac-
[8] O. Saha and R. Dasgupta, “A comprehensive survey of recent trends in
turing, vol. 7, pp. 128–132, June 2019.
cloud robotics architectures and applications,” Robotics, vol. 7, 08 2018.
[33] M. Rüßmann et al., “Industry 4.0: The future of productivity and growth
[9] B. Kehoe, S. Patil, P. Abbeel, and K. Goldberg, “A survey of research in manufacturing industries,” April 2015.
on cloud robotics and automation,” IEEE Transactions on Automation [34] A. Gharaibeh et al., “Smart cities: A survey on data management,
Science and Engineering, vol. 12, pp. 398–409, April 2015. security and enabling technologies,” IEEE Communications Surveys and
[10] “The internet of robotic things,” ABI Research, p. 20, 2014. Tutorials, August 2017.
[11] I. Afanasyev et al., “Towards the internet of robotic things:
Analysis, architecture, components and challenges,” October 2019.
arXiv:1907.03817.
[12] L. Romeo, A. Petitti, R. Marani, and A. Milella, “Internet of robotic
things in smart domains: Applications and challenges,” Sensors, vol. 20,
pp. 33–55, June 2020.
[13] F. M. Anwar and M. B. Srivastava, “Precision time protocol over lr-wpan
and 6lowpan,” in 2017 IEEE International Symposium on Precision
Clock Synchronization for Measurement, Control, and Communication
(ISPCS), pp. 1–6, August 2017.
[14] K. Mekki, E. Bajic, F. Chaxel, and F. Meyer, “Overview of cellular
lpwan technologies for iot deployment: Sigfox, lorawan, and nb-iot,”
in 2018 IEEE International Conference on Pervasive Computing and
Communications Workshops (PerCom ), pp. 197–202, March 2018.
[15] S. Kavuri, D. Moltchanov, A. Ometov, S. Andreev, and Y. Koucheryavy,
“Performance analysis of onshore nb-iot for container tracking during
near-the-shore vessel navigation,” IEEE Internet of Things Journal,
January 2020.
[16] S. Chinchali et al., “Network offloading policies for cloud robotics: a
learning-based approach,” February 2019. arXiv:1902.05703.
[17] S. Hu, F. Rusek, and O. Edfors, “Beyond massive mimo: The potential
of data transmission with large intelligent surfaces,” IEEE Transactions
on Signal Processing, vol. 66, pp. 2746–2758, March 2018.
[18] Y. Yuan, G. Cao, A. Zhu, X. Lyu, and Y. Wang, “Communication scheme
of cloud platform for the lower limb exoskeleton rehabilitation robot,” in
17th International Conference on Ubiquitous Robots (UR), pp. 327–332,
June 2020.
[19] K. Dautenhahn and I. Werry, “Issues of robot-human interaction dynam-
ics in the rehabilitation of children with autism,” May 2000.
[20] M. Nitti, R. Girau, and L. Atzori, “Trustworthiness management in
the social internet of things,” Knowledge and Data Engineering, IEEE
Transactions on, vol. 26, pp. 1253–1266, May 2014.
[21] M. F. R. Lee and T. W. Chien, “Artificial intelligence and internet of
things for robotic disaster response,” in 2020 International Conference
on Advanced Robotics and Intelligent Systems (ARIS), August 2020.
[22] H. Durmuş and E. Gunes, “Integration of the mobile robot and internet
of things to collect data from the agricultural fields,” in 8th International
Conference on Agro-Geoinformatics, pp. 1–5, July 2019.
[23] W. Si, G. Srivastava, Y. Zhang, and L. Jiang, “Green internet of things
application of a medical massage robot with system interruption,” IEEE
Access, September 2019.
[24] T. M. N. U. Akhund et al., “Iot waiter bot: A low cost iot based
multi functioned robot for restaurants,” in 8th International Conference
on Reliability, Infocom Technologies and Optimization, pp. 1174–1178,
October 2020.
[25] P. P. Ray, “Internet of robotic things: Concept, technologies, and
challenges,” IEEE Access, vol. 4, pp. 9489–9500, January 2017.