Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Pires2019 - Digital Twin

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Digital Twin in Industry 4.

0: Technologies,
Applications and Challenges
Flávia Pires∗ , Ana Cachada∗ , José Barbosa∗ , António Paulo Moreira† , Paulo Leitão∗ ,

Research Centre in Digitalization and Intelligent Robotics (CeDRI), Instituto Politécnico de Bragança,
Campus de Santa Apolónia, 5300-253 Bragança, Portugal
Email: {fpires, acachada, jbarbosa, pleitao}@ipb.pt
† INESC TEC - INESC Technology and Science, Faculty of Engineering, University of Porto, Porto, Portugal

Email:amoreira@fe.up.pt

Abstract—The digital transformation that is on-going world- real and virtual systems to collect data, which will impact the
wide, and triggered by the Industry 4.0 initiative, has brought simulation of the model. One of the main capabilities of this
to the surface new concepts and emergent technologies. One of digital copy is to monitor the processes of the physical world
these new concepts is the Digital Twin, which recently started
gaining momentum, and is related to creating a virtual copy of [4]. This concept evolved and transformed into an emergent
the physical system, providing a connection between the real and technology the Digital Twin (DT) [5] that has been gaining
virtual systems to collect and analyze and simulate data in the attention in the last years.
virtual model to improve the performance of the real system. The The emergence of the DT assumes a crucial importance
benefits of using the digital twin approach is attracting significant in the development of Industry 4.0 solutions, as noticed by
attention and interest from research and industry communities
in the last few years, and its importance will increase in the the Gartner hype cycle that places the DT as an innovation
upcoming years. Having this in mind, this paper surveys and dis- trigger of emerging technologies in 2017 [6] and at the peak of
cusses the digital twin concept in the context of the 4th industrial inflated expectations in 2018 [7]. In fact the DT concept falls
revolution, particularly focusing the concept and functionalities, under the trend of digitalised ecosystems, and Gartner expects
the associated technologies, the industrial applications and the that in the next five years, hundreds of million of objects,
research challenges. The applicability of the digital concept is
illustrated by the virtualisation of an UR3 collaborative robot machines or systems will have a DT [7]. Another important
which used the V-REP simulation environment and the Modbus prediction from the Gartner report is that by 2021 most of the
communication protocol. large industrial companies will be using DT, which will result
in a 10% gain in effectiveness [8]. The expectation to have
Keywords: Digital Twin, Industry 4.0, Virtualisation, Simulation. the DT playing such an important role in the years to come,
imposes an upcoming need to increase efforts to make it a
viable solution.
I. I NTRODUCTION
Having in mind the importance that the DT will play
The industrial world is facing a digital transformation that in the upcoming years, the objective of this paper is to
started in Germany, in 2013, with the Industrie 4.0 initiative analyse and survey the DT domain, particularly the concept,
[1]. This transformation, known as the fourth industrial revolu- associated technologies, applications and research challenges.
tion is based on the use of Cyber-Physical Systems (CPS) and The applicability of the concept is illustrated in this paper by a
Information and Communication Technologies (ICT) in manu- small experimental implementation related to the virtualisation
facturing systems, particularly, Artificial Intelligence (AI) and of a UR3 collaborative robot which used the V-REP simulation
Internet-of-Things (IoT). The digital transformation that the environment and the Modbus communication protocol.
manufacturing world is facing implies a transition of the cur- The rest of the paper is organised as follows, Section II
rent industrial paradigm, by transforming the current systems presents an overview of the DT, mainly its origins, evolution
into more flexible and adaptable ones, with the ability to learn and various definitions throughout the time. Section III analy-
and make autonomous decisions [1]. CPS act as the backbone ses the main technologies linked to the DT concept. Section IV
to implement the Industry 4.0 principles and support the discusses several applications of the DT approach in different
creation of a network of manufacturing components with cyber domains. Section V presents the virtualisation of an UR3
and physical counterparts, that have the ability of performing collaborative robot case study under the DT concept. Section
decentralised and autonomous decisions [2], [3]. VI discusses the challenges associated with the DT field.
Industry 4.0 comprises several key design principles, serv- Finally, Section VII rounds up the paper with the conclusions
ing as guidelines for its implementation, namely decentrali- and presents the future work to be developed.
sation, interoperability, virtualisation, real-time capability, ser-
vice orientation and modularity. The virtualisation concept can II. C ONCEPT OF A D IGITAL T WIN
be defined as the capability of the CPS to create a virtual copy The definition of DT has evolved through the years, but
of the physical system, providing a connection between the with the emergence of Industry 4.0 it became more general.

l-))) 
A literature search was conducted to access the evolution of manufacturing sector, which is seen as “the coupled model
the DT along the recent years, using the Scopus API to search of the real machine that operates in the cloud platform and
"Digital Twin" in the 2011 - 2018 time interval, covering two simulates the health condition with an integrated knowledge
years before the launch of Industry 4.0. The search focused from both data driven analytical algorithms as well as other
on the appearance of the DT term in the title, abstract and available physical knowledge” [11]. A broader definition of
keywords of papers indexed by Scopus. Figure 1 shows the DT was provided by [12] where DT is defined as the digital
evolution of the yearly number of publications related to DT. product counterpart of a physical product. This definition
allowed to stretch the fields of application for DT, transforming
it into a more general technology. Fraunhofer IPK and TU
Berlin propose another definition, in which DT is a “digital
representation of a unique asset (product, machine, service,
product service system or other intangible asset), that alters
its properties, condition and behaviour by means of models,
information and data” [13].
The conceptual architecture proposed by [14] for under-
standing how the DT works, illustrated in Figure 2, comprises
a six-step approach that includes the Create, Communication,
Aggregation, Analyse, Insight and Act steps.

Figure 1: Evolution of the number of DT publications.

The Scopus search shows that the publications about DT


started a bit earlier than the appearance of Industry 4.0. With
the launching of Industry 4.0, in 2013, this technology was a
bit left behind and the publication number has reduced, but
in 2016 a boost happened starting to grow the interest in DT,
being expected that by 2019 it keeps growing.
Although the publications with the term "Digital Twin"
have not started until 2010, the origins of the concept dates
back to the beginning of the 21st century, namely in 2002,
where Michael Grieves made a presentation of a “Conceptual
Ideal for Product Lifecycle Management”. Although it was
not called DT at the time, but as “Mirrored Spaces Models”, Figure 2: DT functionality diagram (based on [14]).
common features were presented, such as the real space, the
virtual space and their connection through the data flow from In the Create step, the technologies, as sensors, are imple-
real-virtual-real path. In 2006, the concept started to be named mented in the physical asset to collect data from processes
as “Information Mirroring Model” and by 2010 the term DT and environment. Information about Computer Aided Design
was attached to the concept [9]. According to [9], DT can be (CAD) models and Manufacturing Execution Systems (MES)
defined as “a set of virtual information constructs that fully also passed along. In the Communication phase, adequate
describes a potential or actual physical manufactured product communication protocols are established to send the infor-
from the micro atomic level to the macro geometrical level. mation to the twin. This step comprises three components:
At its optimum, any information that could be obtained from edge processing, communication interfaces and edge security.
inspecting a physical manufactured product can be obtained The first component is responsible for processing data near
from its Digital Twin.”. its source, the second component makes the communication
Another definition was provided by the US National Aero- between the sensors and the integration, and the third compo-
nautics and Space Administration (NASA) that defines the nent keeps the connections and the data safe. The Aggregation
DT as “an integrated multi-physics, multi-scale, probabilistic step is responsible for gathering all the processed data into a
simulation of a vehicle or system that uses the best available repository, making it available for visualisation and analysis,
physical models, sensor updates, fleet history, etc., to mirror which is performed in the Analyse step. The Insight step
the life of its flying twin. It is ultra-realistic and may consider is related to present the knowledge generated from the data
one or more important and interdependent vehicle systems” analysis. Finally, in the Act step, the knowledge created in the
[10]. In 2013, the first steps in applying the concept into previous steps is forwarded to the actuators in the physical
different areas were made, e.g. implementing DT in the asset [14].


The different definitions presented in the literature, led Table I: Technologies and tools for DT.
to a general definition that states the DT as the digital Technologies Tools
copy of a physical object or system, that is connected and
Model Agent-Based Simulation, Anylogic, Simio, Arena, Au-
shares functional and/or operational data. The real-time and Discrete Event Simula- toMod, Enterprise Dynam-
historical data will be used for assessment of the conditions tion, System Dynamics, ics, FlexSim, SolidWorks,
of the physical asset, to perform optimisation and prediction Petri-Nets, CAD AutoCAD, FreeCAD, Iron-
CAD
through the use of machine learning algorithms and simulation
techniques. These outputs will be used to improve the physical Data Acqui- Smart sensors, OPC-UA, Arduino, Eclipse IoT
sition MQTT, TCP-IP, Modbus Project, Node-RED,
asset performance. Kinoma, ModbusExaminer,
IEC Server
III. T ECHNOLOGIES A SSOCIATED TO D IGITAL T WIN
Data Artificial Intelligence, Support Vector Machine, K-
The complexity behind the DT is translated into the "amal- Analytics Machine Learning, Deep Nearest Neighbours, Con-
gama" of required technologies that is required to implement Learning volutional Neural Networks,
Recurrent Neural Networks,
the concept. Figure 3 shows the Industry 4.0 technologies that TensorFlow, Keras
are most found in the retrieved data set of DT publications.
Data Cloud, Edge, Fog Cloud Dataflow, Predix Edge
Processing

basis (e.g., Petri-Nets), or to create a visual model (e.g., CAD


model), which posteriorly can be simulated. According to the
literature review performed by [15] one of the most common
technologies found in the literature for model construction is
discrete event modelling. For the data acquisition, depending
of the system or machine for which the twin is being created,
it is possible to use smart sensors and/or communication
protocols to support the connectivity of the data. The authors in
[15] identified OPC-UA and MQTT communication protocols
Figure 3: Industry 4.0 technologies associated to DT. as two of the most used for data acquisition. In terms of data
analytics, the usage of AI, ML or DL algorithms are the most
Naturally, CPS is the concept that is most found in the common technologies currently. Finally, cloud, edge and fog
dataset, followed by AI, including Machine Learning (ML) computing technologies can be used for the data processing.
and Deep Learning (DL), Virtual Reality (VR) and Augmented
Reality (AR), Big Data and IoT technologies. Considering the IV. A PPLICATIONS AND B ENEFITS OF D IGITAL T WIN
six-step approach, mentioned earlier, the connection between The DT concept started to be applied in several areas, such
the physical and virtual worlds can be performed by using as manufacturing, smart cities, healthcare and retail, providing
IoT technologies, and since the collected data will be in large different functionalities (see Figure 4). DT has also been
amounts, the use of Big Data technology will be also crucial. studied in the areas of shipping oil and gas, constructions and
Simulation and AI, and particularly ML, are crucial to support agriculture.
the data analysis, essentially to support monitoring, diagnosis, The application of DT in the manufacturing sector im-
prediction and optimisation. In this context, a significant pacts the way the products are designed, manufactured and
adoption of DL is taking place, in which the algorithms used maintained. On a high level, the DT can evaluate the pro-
for data analysis can be performed in the cloud. The interaction duction decisions, access the product performance, command
with the user, which is a crucial issue in this field, takes and reconfigure machines remotely, handle the troubleshoot
advantage of the VR and/or AR technologies. equipment remotely and connect systems/processes to improve
In another perspective, the technologies and tools used in monitoring and optimise their control [16]. As example, the
DT are not only those emerged from Industry 4.0, and are CNH Industrial company implemented a DT for a production
strongly dependent of its applications [15]. Aiming to under- line that combines simulation with the analysis of collected
stand this phenomenon, a literature review was performed, data. This DT was responsible for establishing new and more
being the papers classified in type (review, case-study and efficient operations in the company [17]. The DT can also
concept), level of integration of the DT, the area of application be applied for process control, process monitoring, predictive
and the technologies that are used [15]. Table I summarises maintenance, operator training, product development, decision
the technologies and tools for the model, data acquisition, data support, real-time analytics and behaviour simulation. To
analytics and data processing elements of a DT. perform the process control, the DT uses the real-time and
For each DT element there is different types of technologies historical data to feed the virtual decision support system that
and tools that can be used. In the case of the model creation, will help the user to take strategic or operational decisions
it is possible to create a model that has a formal mathematical or implement directly the adjustment in the system operation


prototype. In a futuristic perspective, it would be ideal to have
a DT of a patient, which will provide the physician with the
support on the diagnosis and treatment actions [23].
The application of DT in the retail sector is still very
recent, but offers a great potential in terms of consumer
experience and marketing. The DT can be implemented for
creating virtual models of the customers and modelling fashion
for them, for creating virtual copies of the products and for
identifying the consumer patterns of interest [24].
As referred, the DT can be applied in a variety of sectors,
but in manufacturing this technology will be indispensable to
reach the factories of the future, taking advantage not only
of the reduction of costs, but also in reducing resources,
increasing productivity and effectiveness in the design of new
products, and integrating faster the human operators.

Figure 4: Areas of application and functionalities of DT. V. C ASE S TUDY - V IRTUALISATION OF AN UR3 ROBOT
This section presents the virtualisation of an UR3 collabo-
rative robot to illustrate the applicability of the DT concept.
[18]. The DT also performs the remote process monitoring in Having in consideration Table I this experimental implementa-
real-time, as illustrated in [18], [19]. tion will focus on two parts, the model and the data acquisition.
One of the main concerns that has been raised with the The system architecture illustrated in Figure 5 was adopted.
digital transformation, and particularly with the Industry 4.0
initiative, is the role that the human will play in this new
digital world. The use of DT can contribute for the integration
of the human in CPS, particularly focusing on the operator
training for complex operations. In fact, all manual tasks can
be replicated in an in-line DT, which will allow the operators
to improve their skills and confidence before handling the real
system/device, e.g. the real machine [20]. In the case of design
and development of new products, there is a hard path, but
with the design and simulation capabilities of the DT it will
be possible to speed up the process [18], [19].
According to [21] DT can speed up the commissioning pro-
cess, is also responsible for performing machine optimisation,
reducing the risk of project and its able to perform early fault
detection.
The DT in smart cities includes the setup of a virtual replica
of a city, more specifically a digital representation of urban
networks of the city, such as urban power systems. This would
allow to monitor different devices scattered throughout the Figure 5: Architecture of the case study.
city e.g., traffic lights or water pumps. The DT of the city
could be used to develop prevention strategies or to analyse in The architecture consists of three main components, a
advance the impact of the introduction of new infrastructures. physical part, i.e. the UR3 robot, the virtual part, i.e. the model
One example of this city DT is being developed in Singapore, of the robot represented in the V-REP simulation software and
where the virtual model of the city is being created with a custom Java application. The development of a custom Java
support of 3D maps of the city, aiming to increase the application to perform the virtualisation of the UR3 robot is
resillence of the city in unpredictable scenarios [22]. The DT essential, since it accommodates the different communication
will allow to e.g., monitor the environmental conditions of the parts and the transformation of the data before sending it
city or to plan emergency paths under critical situations. to the virtual model. The pseudo-code used to construct this
In the case of the healthcare industry, the DT can be applied application is presented in Algorithm 1.
in two perspectives, one more directed to the medical devices The developed code is able to retrieve the real-time data
and another more related to the patients. Related to the medical from the physical device and send it to the V-REP virtual
devices, the DT will allow to identify maintenance needs model, while the UR3 robot is moving. The first step is related
before they even arise, to remotely monitor operations and to establish communication with the UR3 robot by using the
to perform simulations on the devices before building the first Modbus protocol, and with the V-REP by using the remoteAPI.


Algorithm 1: Java Custom Application VI. C HALLENGES FOR THE I NDUSTRY A DOPTION OF
Objective: Read UR3 data and send them to V-REP D IGITAL T WIN
Define variables for the addresses, connection and storing
There are several challenges to be faced regarding the indus-
parameters;
trial adoption of DT, which can be grouped in three clusters
Establish communication with UR3;
namely, virtual modelling, company structure organisation and
Establish communication with V-REP;
real-time data and synchronisation.
while !quit do
Request data to UR3; In the case of the area of virtual modelling, several chal-
for i < number of addresses to read do lenges still need to be solved, such as the use of AI for
save the received data; creating the virtual models and the validation of the models.
end The lack of existing methodologies and techniques for the
Convert data to target model; validation of multi-stage virtual models is turning out to be a
Make adjustments in the data; major issue. In fact, the virtual models have to be validated
Save processed data; against the real processes on which they were based, including
Pause communication; intrinsic and extrinsic variables [5]. The lack of techniques
Send processed data; and methodologies for validating the virtual models makes
Restart communication; that the process becomes time consuming and hard. With
end the development of AI based models, the challenge lays in
the integration of the current modelling techniques with the
AI. The authors in [19] expected that the integration of the
current models with complex AI based models, could make
The remoteAPI is an communication application that is part of possible to use advanced simulation, operations and analysis.
the V-REP framework. After establishing the communication, This could also enable the creation of human models that can
the application is continuously requesting data to the UR3 include bio-metric and medical data, the business models and
robot, e.g. related to the joints angles. The received data is sophisticated buildings or cities models. The use of AI can
stored and properly transformed in V-REP compliant data (e.g. change drastically the creation of virtual models, and make
the values of angles are originally in miliradian and need to them more adaptive.
be converted into degrees). Additionally, some adjustments in Another major issue is the structure of today companies,
the data are needed, since the UR3 robot and the virtual model since they present a very partitioned architecture, making very
did not have the same referential. Figure 6 presents the real difficult to have access to all the knowledge from differ-
robot and the virtual copy moving at the same time. ent sectors or departments. The unification and exchange of
knowledge within the company has become a real challenge.
According to [25] this is due to the fact that most organisations
lack knowledge and skills about digital engineering and the
organisation of the current company structures. This can lead
to difficulties in the creation of a data flow framework required
for the creation of a DT. Also, according to [26], organisations
present themselves fragmented views into various sections,
such as design, engineering and manufacturing, and every
section has its own information about the different systems and
how to handle them. But then again there is almost no sharing
(a) (b)
of information between the different sections. To make use
Figure 6: Real and virtual robot moving at the same time: (a) of the DT it is important to have access to information from
the robot is in the initial position and (b) the robot has moved different structures, making the unification of structures a very
to the target position both in the real world and in V-REP. important topic.
Since the proper execution of a DT requires the real-time
This case study covers three parts of the DT functionality access to produced data, real-time connection and synchronisa-
diagram (see Figure 2) namely, Create, since a virtual model tion is a major challenge. In fact the connection and integration
was built, Communication, as communication was established of the physical and virtual worlds is very difficult, namely due
between the virtual and the real system and Aggregation, since to the physical environment that has particular characteristics,
the retrieved data was processed and saved. such as variability, uncertainty and fuzziness, the different
The developed work was the ground for the development of scales of the physical and virtual spaces, and the data from
a DT for the UR3 robot. The pictures presented in the Figure the different entities that is being continuously generated [27].
6 were taken in sequence, which allows a perception of the By being able to work around these obstacles, the real-time
movement of the real robot and the same movement of the interaction will be easier to maintain. In terms of the real-
virtual robot (running in the PC). time synchronisation, according to [28] the synchronisation


between the virtual and the physical model is crucial, but needs [4] M. Hermann, T. Pentek, and B. Otto, “Design Principles for Industrie
a breakthrough in terms of high-speed transmission technology 4.0 Scenarios: A Literature Review,” 2015.
[5] B. Rodič, “Industry 4.0 and the New Simulation Modelling Paradigm,”
to stop being a problem. vol. 50, no. 3, 2017.
Another challenge is the integration of intelligence within [6] K. Panetta, “Top Trends in the Gartner Hype Cy-
decision support systems and consequently the lack of trust cle for Emerging Technologies 2017,” 2017. [Online].
Available: https://www.gartner.com/smarterwithgartner/top-trends-in-
in these systems for performing autonomous decisions [29]. the-gartner-hype-cycle-for-emerging-technologies-2017/
A possible solution for this cloud be the integration of AI [7] Panetta, “5 Trends Emerge in the Gartner Hype Cycle
techniques in the decision support system and to increase the for Emerging Technologies, 2018,” 2018. [Online]. Avail-
able: https://www.gartner.com/smarterwithgartner/5-trends-emerge-in-
trust combine this with simulation techniques. gartner-hype-cycle-for-emerging-technologies-2018/
The identified challenges can be transformed into research [8] C. Pettey, “Prepare for the Impact of Digital Twins,” 2017. [Online].
opportunities, contributing to develop more suitable DT solu- Available: https://www.gartner.com/smarterwithgartner/prepare-for-the-
impact-of-digital-twins/
tions, and consequently for a wider industrial adoption. [9] M. Shafto, M. Conroy, R. Doyle, E. Glaessgen, C. Kemp, J. LeMoigne,
and L. Wang, “DRAFT Modeling, Simulation, Information Technol-
VII. C ONCLUSIONS A ND F UTURE W ORK ogy&Processing Roadmap,” 2010.
[10] E. H. Glaessgen and D. S. Stargel, “The Digital Twin Paradigm for
The launching of the Industry 4.0 program contributed to Future NASA and U. S. Air Force Vehicles,” in 53rd Structures,
a momentum in the adoption of the DT, being one of the Structural Dynamics, and Materials Conference, 2012, pp. 1–14.
[11] J. Lee, E. Lapira, S. Yang, and H.-a. Kao, “Predictive Manufacturing
digital transformation technologies of the upcoming years. System – Trends of Next-Generation Production Systems,” Society of
This technology has evolved from the simple virtualisation into Manufacturing Engineers (SME), no. March 2015, 2013.
a complex digital copy of real assets, being characterised by [12] J. Ríos, J. C. Hernandez-matias, M. Oliva, and F. Mas, “Product Avatar
as Digital Counterpart of a Physical Individual Product: Literature
the combination of several technologies, such as optimisation, Review and Implications in an Aircraft,” in Transdisciplinary Lifecycle
real-time data and machine learning. Analysis of Systems, 2015, pp. 657–666.
This paper presents an overview of the DT concept, namely [13] R. Stark, T. D. If, S. Kind, and S. Neumeyer, “Innovations in digital
modelling for next generation manufacturing system design,” CIRP
its origins and definitions, the technologies that can be used Annals - Manufacturing Technology, no. 1, pp. 169–172.
to construct a DT and the possible applications and research [14] A. Parrott and W. Lane, “Industry 4.0 and the digital twin,” Deloitte
fields in which can be applied. In this case the focus was University Press, pp. 1–17.
[15] W. Kritzinger, M. Karner, G. Traar, J. Henjes, and W. Sihn, “Digital Twin
the manufacturing sector, in which the implementation of DT in manufacturing: A categorical literature review and classification,” in
can bring significant benefits, such as real-time monitoring, 16th IFAC INCOM, 2018.
decision-support based on real data, simulation/optimisation [16] L. Kitain, “Digital Twin — The New age of Manufacturing,” 2018.
[Online]. Available: https://medium.com/datadriveninvestor/digital-twin-
of various operating scenarios and reduction of costs by the-new-age-of-manufacturing-d964eeba3313
saving resources. An important conclusion from the survey [17] ANYLOGIC, “An Introduction to Digital Twin Development,” 2018.
of the DT applications is that currently, very few applications [18] H. Value, M. Catapult, and V. R. Forum, “Feasibility of an immersive
digital twin,” Tech. Rep., 2018.
are reported in the literature. This allows to identify a set [19] D. W. Cearley, B. Burke, S. Searle, and M. J. Walker, “Top 10 Strategic
of challenges for a wider industrial adoption. Aiming to Technology Trends for 2018,” 2017.
illustrate some DT principles, an experimental implementation [20] P. Goossens, “Industry 4.0 and the Power of the
Digital Twin,” Tech. Rep., 2017. [Online]. Avail-
of the virtualisation of a UR3 robot was performed. The able: https://www.linkedin.com/pulse/industry-40-power-digital-twin-
virtualisation of the UR3 robot was performed by developing paul-goossens/
a custom application in Java, where it was established the [21] A. Loetzsch, “Digital twins speed up the commissioning process,”
Control Engineering Europe, pp. 8–9, 2019.
communication between the real robot and the virtual model [22] A. Palau, “Digital Twins in cities: A Virtual Replica of Urban Networks,”
in the V-REP simulation platform. The communication was 2018. [Online]. Available: http://www.resccue.eu/blog/digital-twins-
performed by using the Modbus communication protocol and cities-virtual-replica-urban-networks
[23] H. van Houten, “The rise of the digital twin: how healthcare
the remoteAPI, a functionality provided by the V-REP. The can benefit,” 2018. [Online]. Available: https://www.philips.com/a-
developed experimental work was the ground work in the w/about/news/archive/blogs/innovation-matters/20180830-the-rise-of-
development of the DT and to other future work, such as a the-digital-twin-how-healthcare-can-benefit.html
[24] C. Miskinis, “What Is The Potential of Digital
monitoring, optimisation and decision support system, running Twin Technology In Retail?” 2018. [Online]. Available:
in the virtual robotic model and influencing the physical one. https://www.challenge.org/resources/digital-twins-in-retail/
[25] D. H. Rhodes, “Digital Twins: do you have ghosts in your portfolio?”
R EFERENCES pp. 1–28, 2017.
[26] M. Grieves, “Digital Twin: Mitigating Unpredictable, Undesirable Emer-
[1] H. Kagermann, W. Wahlster, and J. Helbig, “Securing the future of gent Behavior in Complex Systems,” 2017.
German manufacturing industry: Recommendations for implementing [27] F. Tao and M. Zhang, “Digital Twin Shop-Floor: A New Shop-Floor
the strategic initiative INDUSTRIE 4.0,” German National Academy of Paradigm Towards Smart Manufacturing,” IEEE Access, vol. 5, pp.
Science and Engineering (ACATECH), Tech. Rep., 2013. 20 418–20 427, 2017.
[2] E. A. Lee, “Cyber Physical Systems: Design Challenges,” in Proc. of the [28] F. Tao, J. Cheng, Q. Qi, M. Zhang, H. Zhang, and F. Sui, “Digital twin-
11th IEEE International Symposium on Object and Component-Oriented driven product design, manufacturing and service with big data,” IJAMT,
Real-Time Distributed Computing (ISORC’08), 2008, p. 363–369. pp. 3563–3576, 2018.
[3] P. Leitão, S. Karnouskos, and A. W. Colombo, “Industrial Automation [29] D. Daas, T. Hurkmans, S. Overbeek, and H. Bouwman, “Developing
based on Cyber-Physical Systems Technologies: Prototype Implementa- a decision support system for business model design,” no. July 2016,
tions and Challenges,” Computers in Industry, vol. 81, pp. 11–25, Sep. 2012.
2016.



You might also like