See discussions, stats, and author profiles for this publication at: [Link]

net/publication/348480466

Deploying Cobots in Collaborative Systems: Major Considerations and

Productivity Analysis

Article in International Journal of Production Research · January 2021

DOI: 10.1080/00207543.2020.1870758

CITATIONS READS

54 2,195

4 authors:

Yuval Cohen Shraga Shoval

Afeka Tel-Aviv Academic College of Engineering Ariel University
159 PUBLICATIONS 1,636 CITATIONS 187 PUBLICATIONS 1,812 CITATIONS

SEE PROFILE SEE PROFILE

Maurizio Faccio Riccardo Minto

University of Padova University of Padova
152 PUBLICATIONS 3,955 CITATIONS 21 PUBLICATIONS 531 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Yuval Cohen on 14 January 2021.

The user has requested enhancement of the downloaded file.

Deploying Cobots in Collaborative Systems: Major Considerations and
Productivity Analysis

Collaborative robots (cobots) are important components of the Industry 4.0

paradigm and smart manufacturing. Cobots are known for their ability to interact
with the operators in a shared workspace. Due to their spread in the last decade,
cobot research proliferated. However, most individual studies focused on specific
aspects of cobot deployment, and only scant attention was given to their
evaluation (mostly not based on productivity criteria). Thus, better support is
needed for cobot acquisition and deployment decisions. This paper answers this
need by presenting a summary of the major considerations related to cobots
acquisition and deployment, and providing a productivity analysis procedure that
supports cobot acquisition and deployment decisions. Defining the cobots'
required characteristics and capabilities, effectively narrows the possible
selection of cobots. However, it does not give information as to the economic
value of acquiring and deploying a specific cobot. So, in addition to cobots'
characteristics and capabilities, this paper presents computational techniques to
analyse and support this decision for a single workstation per se, and for a station
in an assembly line. The difference between these two cases is discussed and
analysed, and corresponding models are presented with computational examples.

Keywords: cobot, Industry 4.0, collaborative robot, productivity, cobot

acquisition, cobot selection, robot-human interaction

1. Introduction

The fourth industrial revolution known as Industry 4.0 (I-4.0) is related to the

integration of advanced digital technologies and artificial intelligence (AI) in many shop

floors. I-4.0 principles blur boundaries between the cyber and the physical components

as well as the boundaries between humans and non-human components. (Bortolini et al.

2017; Frank et al. 2019; Moeuf et al. 2020; Jiao et al. 2020). However, traditional

industrial robots are designed to operate in isolation from humans, with minimal

physical interaction between them (Arviv et al. 2016). This contradiction motivated

1
designers to develop the new generation of collaborative robots (cobots) that are able to

share the workspace with human operators (Marvel and Norcross 2017; Kadir et al.

2018; El-Zaatari et al. 2019; Hashemi-Petroodi 2020). Thus, cobots became important

component of I-4.0 (Bragança et al. 2019; Cohen et al. 2019a; 2019b; Sherwani et al.

2020; Ivanov et al. 2020). Faccio et al. (2019) elaborated on the differences between

robots and cobots. The major differences between traditional robots and cobots are

summarized in Table 1.

Table 1. The major differences between robots and cobots

Characteristic Industrial Robots Cobots
Role Replacing a worker Assisting a worker
Human interaction Commands, and Intelligent interaction:
programming assigning gesture recognition, speech
locations movements and recognition, and anticipating
gripping. operator moves.
Camera and Computer External camera, and Built -in standard (as part of
external system when exists. the cobot), coupled with
Vision
artificial intelligence.
Workspace Separate safe workspace for Sharing the same workspace.
robots and operators. No fencing.
Usually fenced.
Work envelop Essential and rigid Not relevant
Rapid handling of Usually needs a full setup Built-in standard
disruptions and after disruption.
obstruction
Re-programming Rare Frequent
Physical disruptions Mostly hazardous. Setup Safe with easy re-initiation.
required for re-initiation.
System self-awareness Basic failure detection Real-time monitoring of: (1)
load on each axis and
segment, (2) tactile pressure.,
(3) axis locations.
Agility Rapid motions Slow motions
Payload May be heavy Not heavy
Acquisition cost High Low
Ability to work in No Yes
dynamic environment,
possibly with moving
entities

2
Cobots safely share the same workspace with humans, and are known to be very safe.

For example, following extensive use of cobots in Carlsberg corporation, Carlsberg

reported in 2019, a significant reduction in accidents at its main plant in Frederica,

Denmark (URL: [Link]

fredericia-cuts-factory-floor-accidents-with-cobots/; retrieved November 2020). The use

of cobots not only significantly improved employee safety, but it brought the plant

closer to zero accidents. While cobot literature proliferated, only scant attention was

given to the cobot acquisition and deployment decisions (Fast-Berglund et al. 2016;

Cohen et al. 2019b; Simões et al. 2020). The aim of this paper, is therefore, to facilitate

the acquisition and deployment decision-making process. This aim is achieved by

attaining the following goals:

1. Present a comprehensive literary coverage of the various considerations related

to cobot deployment in assembly systems. This is achieved in sections 2.

2. Present cobot characteristics and capabilities. This is achieved in section 3.

3. Present productivity analysis of cobot deployment (sections 4, 5). Cobot

integration into a single workstation is described in section 4, and the case of

assembly line is presented in section 5.

Cobots may be integrated to variety of manual workstations performing various

activities such as: picking, packing, assembling, palletizing, welding, handling material,

inspecting parts and products (Vojić, 2020), loading/unloading machines, part cleaning,

bin picking, and kitting (Antonelli and Bruno, 2019). The most popular environment for

these cobots' activities is related to assembly environment (Fast-Berglund et al. 2016;

Malik and Bilberg 2017, 2019; El Makrini et al. 2018; Bejarano et al. 2019; De Winter

et al. 2019; Antonelli and Bruno 2019; Cohen et al. 2019a, 2019b; and Guo et al. 2020).

3
 Cobots integration in industry has been growing (Marvel and Norcross 2017). In

2016, the US market share of cobots reached only $100 million and grew to $ 710

million in 2018. (Djuric et al. 2016). According to Markets and Markets research report

(2019), US forecast for 2025 is $12,303 million due to fast growth in the use of cobots

in assembly applications. It is expected that standard interfaces will further improve the

scalability and accessibility of cobots (Cohen et al. 2019b).

The decision on the acquisition and deployment of cobots is a complex

productivity decision (Fast-Berglund et al. 2016; Simões et al. 2020). It involves the

timing of purchase, lifetime expectancy, the selection of work-stations for deploying the

cobot, the selection of cobot type, and even the social and cultural implications of

deploying cobots in industry (Romero et al. 2016). The benefits vs. costs trade-off of I-

4.0, and in particular collaborative robots, still remains to be researched (Fast-Berglund

et al. 2016; de Man and Strandhagen, 2017). Initial step of presenting productivity

considerations related to cobots' deployment was taken by Cohen et al. (2019b). A

summary of the variety of considerations related to cobot deployment is presented in the

literature is presented in Table 2.

The rest of the paper is structured as follows: section 2 analyses the related

literature, identifying the various cobot deployment considerations; section 3 focuses on

the cobot selection considerations; section 4 analyses the effect on productivity of a

single cobot's deployment in a manual system. Section 4 defines the notations for the

proposed models, continues with the model development and its analysis, and a

computational example, that demonstrates the attractiveness of using cobot

. Section 5 present a model for the case of cobot deployment in assembly line,

followed by numerical computation. Finally, section 6 concludes the paper.

4
 2. Literature review

The deployment of collaborative robots is a complex problem whose different aspects

have been identified in the literature in recent years (Fast-Berglund et al. 2016; Simões

et al. 2020). Indeed, the cobots’ capabilities allow them to safely share the workspace

with human operators, to be quickly reprogrammed, and easily applied in several

industrial scenarios, motivating the study of their optimal deployment (Sherwani et al.

2020). New technologies such as augmented reality (AR) and mixed reality (MR) have

potential to facilitate this collaboration (Ostanin et al. 2020)

The framework presented in (Djuric et al. 2016), is a tool to identify the

different challenges to guide the deployment of cobots. Another approach is presented

in Malik and Bilberg (2019): the proposed evaluation method identifies product and

process characteristics and rates each one. The methodology was applied to an industrial

case study, showing that cobots deployment resulted in the automation of up to 70% of

the workload.

Gil-Vilda et al. (2017) analysed deployment of cobots by using a real case study:

they conclude that cobots are very different in their roles and operation from traditional

robots due to their nature, and the following capabilities:

• Mobility (M), i.e. the ability to easily move the cobot in the production plant.

• Intelligence (I), i.e. the awareness of the resources and job characteristics, and

their implications.

• Connectivity (C), i.e. human-cobot communication, and cobot system

communication

• Actuation (A), i.e. the ability to develop safe and dynamic trajectories.

5
 • Human centred (HC), the support to the human operator from the physical,

mental, psychosocial point of view.

The letters in the brackets of this list are used in the capability column of Table 2.

While most cobot implementation case studies focus on subsets of these

capability as shown in Table 2, several case studies focus mostly on one of these

capabilities. For example, Bogner et al. (2018) and Liu et al. (2019) focused on the

intelligence, while Zhong et al. (2013) and Guo et al. (2020) focused on connectivity

and actuation)

While reviewing the literature we found 7 major aspects that most cobot papers

could be categorised to: 1. Cobot business model, 2. Cobot deployment framework, 3.

Cobot collaboration, 4. Task assignment: cobot vs human, 5. Cobot safety, 6. Social

perspective: cobots as part of automation revolution, 7. Human-centric perspective:

cobots as part of I-4.0

Table 2 summarizes the recent literature in these major aspects related to cobot

deployment. For each paper, Table 2 highlights the adopted method, the objective, the

application considered in the paper, and the aforementioned cobot capabilities.

In Table 2 the most adopted methods are frameworks, mathematical models, and

surveys, with the main aim of improving the cobot operation in various ways. The most

popular application is assembly and the most considered capability, according to the

authors’ classifications, is the human-robot communication.

The studies presented so far, highlight the need to present all the cobot capabilities and

their influence on the deployment of collaborative robots on the system's productivity.

Indeed, despite of their capabilities many collaborative robots are used for machine

tending tasks (Michaelis et al. 2020), using all their capabilities can greatly affect the

decisions made when designing collaborative work cells.

6
Table 2. Examples of cobot deployment aspects considered in the literature
Cobot
Capabilities
Deployment References Method Objective Application
Considered
Aspect
Fast-Berglund Mathematical
Improve flexibility Assembly M
2016 model
Sadik and Urban
Framework Improve agility Assembly HC, C
Cobot 2017
business Analytical
Mecheri and
model Hierarchical Select suitable cobot - HC, I, M
Greene, 2019
Process - AHP
Mathematical
Fager et al 2020 Improve efficiency Kitting A
model
Identification of design
Djuric et al. 2016 Framework Adv. Mfg. M, C, HC
challenges
Bortolini et al. Impact of I4.0 on
Cobot Framework Assembly C, I
2017 assembly systems
deployment
De Winter et al. Learning task from the
framework Algorithm Assembly C, I
2019 operator
Bortolini et al. Motion Production/ergonomics
Adv. Mfg. M
2020 Analysis assessment
Tamas and Murar Experimental
Improve flexibility Adv. Mfg M, C, I
2018 method
El Zaatari et al. Mathematical Evaluate fluency in
Assembly M, C
2019 model HRC
Cobot Improve makespan and Packaging
Pearce et al. 2018 Algorithm A, HC
collabo- ergonomics & assembly
ration Hoffman, 2019 Analysis Fluency of interaction I
Matheson et al. Cobot collaboration M, A, I, C,
Review Adv. Mfg.
2019 overview HC
Bejarano et al.
Case study Effective collaboration Assembly I, A, HC
2019
Malik and
Framework Improve productivity Assembly C
Task Bilberg, 2017
El Makrini et al.
assignment: Case study Test feasibility Assembly HC, C, I
2018
cobot
Malik and
vs Algorithm Task allocation Assembly A
Bilberg, 2019
human Faccio et al. 2019, Evaluate product
Algorithm Assembly A
2020. influence on HRC
Common safety
Vicentini, 2020 Survey - A, I
terminology
preloaded Impact mitigation in
Cobot Seriani et al. 2018 A
control cobots
safety Novel Safety Avoid human robot
Ayoubi et al. 2019 A
Mechanism collision
Heo et al. 2019 Deep learning Avoid cobot collision Adv. Mfg A
Calzavara (et al. Cobot help for ageing
Survey Adv. Mfg HC
2020) workers
Cobot Hashemi-Petroodi Cobot for helping in
Survey - HC
impact on et al. 2020) peak demand
workers Impact of technophoby
McClure, 2018 Survey - HC
and society on financial unsecurity
Addressing fear and
Fiaidhi et al. 2018 Survey Industry 4.0 HC
courage of technology
Romero et al. Explore technologies
Review Industry 4.0 HC, C, A
2016; and HRC
Human- Empirical Identify human
Kildal et al. 2018 Adv. Mfg C, HC
analysis concerns
centric
Gustavsson et al. Select communication
perspective Analysis Adv. Mfg C, HC
2018 technologies
Pollak et al. 2020 Experimental Cobot related stress Adv. Mfg HC
Michaelis et al.
Analysis Improve interaction Adv. Mfg HC, I
2020

7
 While this paper presents new models for economic justification for cobot

deployment, cobot deployment could be also justified based on other considerations

such as ergonomic considerations, load and fatigue reduction, the well-being of the

operator, and in some cases even quality. These goals can be achieved in different ways.

Therefore, a more detailed review of the cobots' characteristics and capabilities is

presented in section 3, along with the technology that allows to adopt them). Moreover,

most expected benefits of cobots are related to their collaborative nature. The

collaborative nature of cobots is related to the human factors. So, capabilities related to

human factors will be stressed in the summary of cobots' characteristics and capabilities

presented next.

3. Contemporary cobot characteristics and capabilities

3.1 Safety Standards for Industrial Cobots

Cobot technology is not fully mature, and is still evolving. So, cobot standardization is

in fledgling stages. Some cobots standards exist, and some relevant robotics standards

are adopted for cobot safe and successful implementation (Vicentini, 2020).

The main international cobot standard is ISO/TS 15066: a specifications of

safety requirements for collaborative robot applications. However, ISO/TS 15066 does

not stand alone, and as Figure 1 shows, it is part of larger ISO family of standards.

ISO 11161 – Integrated Manufacturing System

ISO 10218-2 ISO/TS 15066

Industrial Robots Cobots
Collaborative Robots

Figure 1 Cobot standard ISO/TS 15066 as part of ISO 11161 family of standards

8
In particular the ISO/TS 15066 specifies four modes of operation:

1) Safety-rated monitored stop (SMS) - In this mode, whenever a human or other object

is in close proximity, to the cobot stops its motion (without shutting the engines down).

2) Hand-guiding (HG) - In this mode, the operator uses a manual guiding device in

proximity to the robot end-effector, to guide the cobot's motion.

3) Speed and separation monitoring (SSM) - In this mode, the robot continuously keeps

a minimal safety separating distance between robot and operator. When the distance is

wider than the separating distance, the cobot the and operator may move concurrently in

the shared workspace. If the separating distance is breached, the cobot slows to a halt.

The cobot resumes operation when the operator moves away.

4) Power and force limiting (PFL) In this mode, the robot system may have closely

controlled contact with the operator (or its surrounding) as long as it is under threshold

limit values for movement (almost static), and pressure. Above these values the cobot

is released and is pushed by the operator (or other external force).

Cobots are also part of the general scope of machinery safety, regulated by the

following reference standards:

• UNI ENISO 12100:2010 “Machine safety, general design principles, risk

assessment, and risk reduction”.

• UNI EN ISO 10218-2:2011 “Robots and equipment for robots, Safety

requirements for industrial robots"

9
 • UNI EN ISO 10218-1:2012 “Robots and equipment for robots, Safety

requirements for industrial robots"

3.2 Technical Characteristics of Industrial Cobots

3.2.1 Common cobot characteristics

In this section we describe characteristics that are common to the large majority of

cobots. The external shape of cobots typically has a smooth round appearance, and

absence of sharp edges and rough surfaces. This is essential for safety in the presence of

humans, and prevents clamping situations.

To further reduce the severity of collisions, all cobots are controlled with a

limited power and force, to ensure safe operation in the presence of humans (ISO

15066). This means that all of their axes are monitored for load, velocity, and

acceleration. Each axis has a threshold for the power required to stop it in its current

inertia and acceleration a feature absent in traditional robots.

The axes control is carried out by using built-in force-torque sensors. These

sensors can be installed in each axis, in the flange or in the cobot base, in order to detect

collisions and immediately stop the cobot. Other typical cobot sensors are proximity

sensors and pressure sensors. Proximity sensors continuously control the distance from

each arm segment to external objects (such as the operator). Pressure sensors control the

pressure exerted by the cobot parts on external objects.

Cobots are typically fitted with at least one camera in proximity to their end-

effector tip. This camera is typically used to identify a gripping object, and locate the

end of the cobot's arm in space. There are many other possible uses of a camera, but

they vary between cobots. The reduction in camera costs is expected to enable installing

several cameras on each cobot and increase use of smart computer vision capabilities.

10
3.2.2 Cobot specific features

In this section we describe features that differentiate cobots from one another, and may

help the acquisition decision related to the type of tasks the cobot may perform. The

main categories are: payload, maximal reach, number of axes, repeatability, maximal

speed, force sensing, and special features. These categories are described as follows:

Payload:

The range of cobot activities is limited by its payload. Therefore, before the cobot

acquisition, the range of its possible activities should be analysed to determine the

required maximal payload. Moreover, payload is the main feature by which the cobot

models are characterized. Payload has significant implications related to the force,

torque and required engine power. Although the payload of collaborative robots is

usually limited to weights less than 15 kg to avoid the risk of moving heavy masses,

some six-axis cobots present capability of much higher- payload.

Number of axes:

Cobots typically have extra axes in their arms to enable better manipulation of their

arms around obstacles, and collisions avoidance. Cobots may have various kinematic

schemes, with the most common configurations represented by 6 and 7 axes chains, as

seen in Table 3. Moreover, differently from traditional robotics, adding a redundant

joint, i.e. using a number of axes higher than 6, could allow to avoid imminent

collisions, therefore it is comprehensible that robot manufacturers have introduced 7-

axis cobots.

Maximal reach:

Maximal reach is measured from the centre of the cobot's basis. Thus, the maximal

11
reach limits the radius of actions of the cobot. Typical ranges of cobot's max reach is

between 0.5 to 1.5 meter. Since arm's reach affects the physical load on all cobots axes,

majority of cobots have maximal reach of less than one meter. Only few exceptions

have arm reach above 1.5 meter.

Repeatability:

This is a measure of preciseness. Typically, much less than a millimetre. If the cobot

must reach precise locations and orientations this parameter becomes critical.

Maximum Speed

The highest speed at which the cobot’s end effector can move is the maximum speed.

According to ISO 15066 safety speed must be analysed for each task because even

collaborative robots can go very fast and become unsafe in certain situations.

A short list of other additional important parameters includes: cobot weight,

cobot price, maximal operating temperature, power consumption, IP rating (dust and

liquid resilience). The aforementioned parameters are sufficient for the scope of this

paper.

12
Table 3. A sample of widely used cobots with main characteristics

Manufacturer

Productive
Kawasaki

Universal
Robotics
Comau

KUKA

Robots
AUBO

Fanuc
ABB
i3 CR-4ia LBR iiwa UR3
i5 CR-7ia 7R800 UR5
IRB
Cobot
14000 Aura CR- duAro OB7
Models YuMi i7 UR10
15ia LBR iiwa
CR- 14R820
i10 UR16
35ia
3 5 3
7
5 7 5
Payload [kg] 0.5 170 2 5
7 15 10
14
10 35 16

Joints 7x2 6 6 6 4x2 7 7 6

Repeatability
0.02 0.05 0.1 0.03 0.05 0.15 0.1 0.02
[mm]
625 550 500
800
Reach 924 717 850
559 2800 760 1000
[mm] 1150 1441 1300
820
1350 1813 900
1900 1000
Max speed 2800 1000
1500 500 2000 250 2000 1000
[mm/s] 3500 1500
4000 750
Force
Force Joint Joint Skin Joint Joint Joint Joint
sensor
sensing sensing sensing sensing sensing sensing sensing sensing
at base
Torque
Max
sensors in Load
Dual arm payload Passive Dual Mobile
Features redundant of skin arm
each joint,
base
cell on
mobile flange
category
base

3.3 Cobot capabilities

Cobot's technical characteristics (discussed in section 3.2) can vary widely (Malik and

Bilberg 2019). These characteristics are task related features that describe what the

cobot is capable of doing with its software, hardware, and some peripheral sub-systems

13
(such as external cameras and part feeders). These, task related features are defined here

as cobot's capabilities. Our categorization of main capabilities, is an adjustment and

some additions to the capability model by Gil-Vilda et al. (2017) presented in the

literature review. The additions are supported by case studies of cobot deployment (e.g.,

Liu et al. 2019). Our proposed model is depicted in Figure 2.

Human Support Cap.

Interaction (connectivity) Cap.

Intelligence Cap.

Mobility Cap. Actuation Cap. Processing & status-

tracking Cap.
Safety Capabilities (Cap.) Input Cap.
(Sensory & comm.)

Figure 2. Cobots' main capabilities in rising levels of sophistication

In Figure 2, we added to Gil-Vilda et al. (2017) the following additional capabilities:

"Safety", "input (sensory and communication), and "Processing and status tracking". We
retained the remaining four capabilities: Mobility (M), Intelligence (I), Connectivity (C),
Actuation (A). Finally, to better reflect the meaning of the Human centred (HC)
capability, we renamed it "Human Support" capability.
The addition of the safety dimension is rooted in the fundamental requirement of
sharing common workspace with humans. The importance of safety to cobots is reflected
in the ISO 15066 cobot safety standard (the only standard devoted to cobots). The
addition of input capabilities (related to sensors, cameras, microphones, and electronic
messaging) is also fundamental dimension of cobots. Therefore, "Safety" and "Input"
capabilities are at the basis of cobots capabilities in Figure 2.
On top of the fundamental layer of "Safety" and "Input" capabilities (in Figure .
2), there are the "Mobility", "Actuation" and ""Processing and status tracking"
capabilities. They also support the "Intelligence" capability in the next level up (which
refers to awareness of the resources and job characteristics and their implications).

14
Mobility allows to move the cobot to other parts of the production plant in a faster and
easier way (Gil-Vilda et al. 2017). To improve its flexibility, and also improve the safety,
the cobot actuators must be not only precise, but also capable to dynamically modify their
trajectories.
Lastly, the nature of collaborative robots allows to move towards a more human-
centred design, leading to the next capability level of cobot's ability to improve the human
factors, e.g. learning and forgetting, physical workload, body posture, which should be
considered to improve productivity and the workplace condition (Grosse et al. 2015).
Moreover, the interaction with the operator requires higher level of intelligence, i.e. the
cobot must be aware not only of its status, but also of its environment and the intention
of the human operator. Both human-to-cobot, needed for teaching instructions to the
manipulator, and cobot-to-human, to inform the operator of the cobot intention.
Competence levels for each capability
For each capability, different cobots may have different competence level of that

capability. For example, in actuation: when the operator is close the cobot, the cobot

may act in two different ways: (1) stop, or (2) dynamically change its velocity and

direction. While considering the capabilities, each cobot presents a different

competence profile of the main capabilities. To facilitate the construction of such

profiles, we propose four competence levels for each capability. The minimal

competence level is 1, and the maximum competence value is level 4. This competence

model can be a reference point for future competence or maturity models. The

competence level for each capability are presented in Table 4, and briefly discussed in

the appendix.

4. Productivity analysis for adding a cobot to a manual workstation

So far, the cobot literature has been quite blurred when it comes to the justification for

the acquisition, installation, and deployment of cobots (Cohen et al. 2019b). Relevant

papers only use general qualitative language.

15
Table 4. Cobot capabilities and their suggested competence levels

Cumulative Competence Level

Cobot
Capability
Level 1 Level 2 Level 3 Level 4

ISO 15066 ISO 15066

ISO 15066 ISO 15066
Safety-rated Speed and
Safety monitored stop
Hand-guiding
separation
Power and force
(HG) limiting (PFL)
(SMS) monitoring (SSM)
Cobot sensors:
IoT:
e.g. joint sensors, Vision: one or
Input proximity more cobot
Getting
Actively seeking
**not commands information from
sensors, skin cameras (2D, or missing information
or direct human surrounding
sensing, pressure 3D) and relevant +speech recognition
interface sensors and
and base-force software
cameras
sensor
Cobot follows Cobot decides
Requires cobot to
Rooted in a pre-programmed autonomously to
Mobility single station
be moved between
schedule to move which station to
station
between stations move
Cobot
Cobot stops after
Repeatability dynamically
a collision, and Active co-
Actuation and self-setup
reassesses its new
corrects the
manipulation task
after collision. trajectory to avoid
location
collision
Processing Track operator
Track the cobot
Task status Neighbouring movements and
and Status current tracking. processes tracking gestures (Liu and
tracking kinematic profile
Wang 2018)
Cobot aware of Cobot aware of Cobot aware of Cobot aware of
Intelligence the status of the itself, and of the the status of other human operator
current task surroundings. connected devices status and intention
Cobot current Human activated
Cobot warnings, Real Cobot
state. Real-time menu of possible
Interaction and human responsive to
cobot alarms. cobot's assistance
(connectivity) corrections for the human gestures
Human turns and feedback based
cobot and speech.
cobot ON/OFF on AI
Cobot performs
ergonomically Cobot brings tools Cobot holds and/or
challenging tasks: or parts next to manipulates the
Cobot performs the operator, and
dirty, hot, humid, tool (e.g., screw-
repetitive, and/or takes them away.
and noisy driver, plyers, etc.)
Human dangerous tasks, Cobot issues
environment. or workpiece. May
support And sounds reminders, and
Cobot issues initiate tasks: "let
alarm in draw attention to
safety warnings, me hold it". May
emergency. evolving
and suggests help suggests help in
only in situations. extreme cases.
emergencies.

16
For example, “cobots can also reduce assembly ergonomic complications due to

physical and cognitive loading, while improving safety, quality and productivity”

(Battaïa et al. 2018). However, the work characteristics in different assembly stations

differ greatly, and cobot evaluation is needed to find stations with high improvement

potential (Simões et al. 2020). There is a need for a clear, detailed model for cobot

justification. This section addresses this need. Notations are needed to continue the

analysis and the discussion, and are presented next.

4.1 Notations

The following notations are used in the remainder of the paper.

A– Absenteeism rate (%) is the percentage of absent days per operator. Such
absenteeism case necessitates the employment of an inexperienced
worker at the workstation.
C– Takt time: the standard (nominal) time between each pair of consecutive
products leaving an assembly workstation.
cs – The saved standard time (of the operator) out of C, due to cobot
deployment. (may be different than cobot's moving time).
F– Percent of throughput lost due to fatigue and deterioration
L– Learning loss as (%) percent of time added to the standard nominal time,
due to the learning of inexperienced worker (substituting an experienced
employee, or an absentee).
Nom – The nominal workstation throughput rate per shift without a cobot (e.g.,
Nom=480/C for 480 minutes shift and C in minutes)
CNom – The nominal workstation throughput per shift with cobot (e.g., for 480
minutes shift and C, cs, in minutes: CNom=480/(C-cs))
P– Percent of throughput lost due to performance issues and preciseness
requirements.
Q– Percent of throughput lost due to quality issues
TO – Turnover rate (%) is the percent of workstations with inexperienced
workers (replacing the old ones).

17
 THPT – Real throughput of manual workstation without cobot, during one shift,
with all detrimental considerations included.
C-THPT – Throughput using the cobot during one shift
Lifecost – Lifetime cost of a cobot: purchase, maintenance, and deployment costs
throughout the cobot's life (present value at the acquisition).
Uprofit – the expected profit due to a produced unit/item.
Life – Number of expected cobot deployment years
n– Number of cycle/repetition
tn – Cycle time of the nth cycle repetition
b– Learning coefficient (typical values of b are in the range 0.6>b>0)
- Learning slope (1>>0)
Tn – Total time to complete the first n cycles/repetitions
tstd – Standard time for a given task
NC – Number of cobots deployed in an assembly line

4.2 Analysis of a cobot addition to a manual workstation

For clarity and focus, we limit the analysis to the addition of a cobot to a fully manual

workstation performing routine work. Thus, the model is based on a straight comparison

of productivity of the manual station before and after the cobot deployment. The

throughput of the original station without a cobot is computed as follows:

THPT = Nom*(1-L(A+TO)-P-F-Q) (1)

The throughput of the collaborative station with a cobot is computed as follows:

C-THPT=CNom*(1-L(A+TO)-P-F-Q) (2)

The cobot effect can be measured by percentage of throughput growth:

C-THPT / THPT.

This addition of products could be used to decide whether or not the cobot purchase and

18
deployment costs are justified or not.

Based on (1) and (2), the throughput percent increase due to the cobot is:

C-THPT/THPT = CNom/Nom (3)

For a typical 8 hours shift (480 min.) the throughput percent increase is:

CNom/Nom = [480/(C-cs)]/[480/C] = C/(C-cs) (4)

So the cobot's contribution to the throughput is:

(C/(C-cs)-1)*THPT = (cs/(C-cs))*THPT (5)

Consequently, assuming 250 workdays per year, the cobot acquisition and deployment

is justified as long as the inequality (6) holds true:

[(cs/(C-cs))*THPT ]*250*Life*Uprofit > Lifecost (6)

Note that for simplicity, Net Present Value in (6) is ignored. However, the importance

of Net Present Value of the lifetime costs and the profits, depends on the cobot's

lifetime (Life) and the maintenance costs. The appendix presents Present-Value

calculations, but the paper continues with the simpler computations.

Another important inference from this model is the minimal time that a cobot must save

to justify its acquisition. We use inequality (6) to develop the answer. First step is

isolating cs on the left side of the inequality – as in (7):

cs/(C-cs) > Lifecost/( THPT*250*Life*Uprofit) (7)

For computational simplicity, we name the right side of (7) RHS:

RHS= Lifecost/( THPT*250*Life*Uprofit) (8)

So using (8) on (7) and multiplying both sides by C-cs yields:

cs> (C-cs) RHS (9)

cs [1+ RHS]>/C*RHS (10)

cs > C*RHS / [1+ RHS] (11)

19
Since RHS/(1+RHS)<1 the right hand side of (11) is smaller than C and is the threshold

value of cobot time savings to justify the cobot.

4.3 Example of Cobot Productivity Analysis

The following is an example of comparing) throughput of a single workstation: with and

without a cobot. The example is based on a typical shift of 8 hours (480 minutes) per

workday and 250 annual workdays.

The following is a summary of the example values for throughput related calculations:

Shift=480 minutes, C=1 min. L=6%, A=5%, TO=1%, P=3%, F=2%, Q=1%, Life=5

years, Uprofit= $10, Lifecost=$250,000.

For the example, manual processing time on each product is C=1 minute, and the

deployment of cobot saves only 6 seconds of the 60 seconds manual work, so cs=0.1

minute. From equations (3) and (4) we have:

C-THPT/THPT = C/(C-cs) = 1/(1-0.1) = 1/0.9 = 1.111

So, in this case the cobot deployment resulted in throughput increase of 11.1%: The

throughput of the manual station without a cobot is computed using equation (1): While

the nominal throughput is 480 (Nom = shift/C = 480/1 = 480), the actual throughput

(using equation (1)) is:

THPT = 480*(1-0.05(0.04+0.01)-0.02-0.02-0.01) = 480*(1-0.025-0.05) = 444

The actual throughput increase per shift is given by the right side of equation (5):

(cs/(C-cs))*THPT = 0.1/(1-0.1)*444 = (1/9)*444 = 49

This means that the cobot contributes 49 products per shift!

The overall increased profit due to the cobot is expressed at the left-hand side of

inequality (6): 49*250*5*10 = $ 612,500

20
So, the cobot deployment would be justified only if the cobot lifetime costs are smaller

than $ 612,500. That is, as long as: 612,500>Lifecost

In this example the cobot's lifetime costs are Lifecost=$ 250,000 and since

612,500>250,000 the cobot deployment is justified.

Next, the RHS value from equation (8) is computed for the example:

RHS= Lifecost/( THPT*250*Life*Uprofit) = 250,000 / (444*250*5*10) RHS= 0.045

Using RHS=0.045 and C=1 in inequality (11) yields:

cs > 1*0.045 / [1+ 0.045] → cs > 0.045/1.045 → cs > 0.043

As long as the cobot can save more than 0.043 minutes (2.6 seconds) of each cycle the

cobot lifetime cost of $ 250,000 is justified. Note also that (11) may be written as (12):

cs/C > RHS / [1+ RHS] (12)

The left side of (12) is the cobot savings percentage of the manual takt-time. The right

side of (12) is the minimal required saving (% of takt-time) to justify a cobot

acquisition, and depends only on RHS. The graph of this relation is shown in Figure 4.

cs/C Min. % cycle-time saved by cobot (cs/C) per RHS

0.5 for justifying cobot deployment
0.4

0.3

0.2

0.1

0
0.31
0.01
0.04
0.07
0.1
0.13
0.16
0.19
0.22
0.25
0.28

0.34
0.37
0.4
0.43
0.46
0.49
0.52
0.55
0.58
0.61
0.64
0.67
0.7
0.73
0.76
0.79
0.82
0.85
0.88
0.91
0.94
0.97

RHS= Lifecost/( THPT*250*Life*Uprofit)

Figure 4. Minimal required percent takt-time savings vs. RHS for justifying a cobot
*
RHS= Lifecost/( THPT*250*Life*Uprofit), cs =cobot time savings per takt-time, C = takt-time

21
5. Cobots deployment in assembly lines

5.1 Adding a cobot to a station: Single station vs. a sequential assembly line

While the analysis of section 5 pertains to the addition of cobots to any workstation,

(including assembly lines) long assembly lines have additional vulnerabilities that

require additional analysis. While assembly lines enjoy the advantages of linear

sequential workpiece flow, their throughput is limited by the rate of the bottleneck

station. The bottleneck station is the slowest station in the line. The downstream stations

in the line simply receive the workpiece at the rate of the bottleneck because by

definition (of not being bottleneck) they complete their tasks on the workpiece in a

shorter time and wait until the next workpiece arrives. Figure 5 depicts this situation.

Bottle-neck
Station = i

Station i+1

Figure 5. The pace of the bottleneck station becomes the rate of its successors, and the
line's throughput rate

Manual stations in assembly lines are designed and deliberately assigned work load

below the line's takt-time. Thus, normally they rarely become bottlenecks. However, in

situations of absenteeism and turnover, a new worker is assigned to them, initiating a

learning process that starts with prolonged cycles that get shorter with experience. In the

early stages of such a learning curve the substitute worker has high potential of

becoming a bottleneck for the whole line.

22
 Absenteeism and turnover are encountered daily in a line having dozens of

stations or more (although it is unknown in advance in which station it will occur).

Thus, in situations of absenteeism and turnover, having a cobot that may help the

substitute worker and relieve them from part of the load can improve the whole line's

throughput.

Deployment of substitute workers is the major cause for longer station cycle

times, which slow the entire line. However, deploying cobots in these stations together

with the substitute workers can reduce the cycle-time and keep it close or even below

the takt-time. The effect of deploying a cobot in such a station with a substitute worker

is depicted in Figure 6.

Figure 6. Example of substitution worker’s learning curve before and after cobot
deployment in assembly line.

In Figure 6 the part of the learning curve above the takt-time reflects the period when

the replacement worker is likely to become a bottleneck for the entire line. The area

between the learning curves and the takt-time is the time lost due to the learning curve.

As shown in Figure 6, this area is greatly reduced by the cobot.

23
 However, absenteeism and turnover are random phenomena that are unknown in

advance, and can not be projected exactly. Therefore, the nature of the cobot

deployment for such contingencies is very different. These contingencies occur at

different times in different stations along the assembly line. Tackling this problem

(assisting inexperienced replacement workers) is a dynamic task with shifting locations.

Therefore, installing a static cobot in a given workstation is not an efficient solution.

Instead we propose a scheme that is designed for this situation as described next.

5.2 The proposed scheme for contingency cobot deployment in an assembly line

Since absences and other employee replacements occur randomly at different times in

different stations along the assembly line, the proposed solution should allow changing

cobot station assignment in each shift. This means that the chosen cobot should have

sufficient mobility for quick installation at the beginning of each shift.

In long lines, of several dozens of workers, several worker replacements may

occur simultaneously. In such lines, more than one cobot is usually necessary. The

extreme solution would be to dedicate a cobot for each manual station. However, this is

inefficient and costly. Therefore, we propose dedicating a cobot for groups of manual

stations or line segments.

The following are the major steps proposed here for this purpose.

1. Developing a contingency plan for each manual workstation for assisting

substitute workers by using a cobot.

2. Segmenting the assembly line to segments or small sets of manual stations (each

segment contains 5 to 10 stations). The main cobot features (such as weight and

repeatability) for a given set of stations, should not substantially differ.

24
 3. Dedicating a cobot to each line-segment (or small set of manual workstations).

The cobot of a given segment has to fit well with all the contingency plans for

the segment's stations developed in step 1.

4. Writing the code for the chosen cobot to perform the contingency plan (step 1)

in each manual workstation.

5. Testing the code for each station with an inexperienced worker, and fixing bugs.

5.3 Throughput-cost of replacing absent experienced worker in assembly line

In this section we discuss the effect of an absent experienced worker and his/her

replacement in an assembly line. In lines with many workstations, turnover and

absenteeism are daily occurrence. So, typically in a substitute worker is found in at least

one of the stations along the line (Frick et al. 2018). A replacement of an experienced

worker by inexperienced one, results in a learning curve (Bukchin and Cohen 2013;

Peltokorpi and Niemi 2019, Glock et al. 2019). The first cycle-times in learning are

excessively long. In a sequential assembly line, the initial stage of such learning curve is

a potential bottleneck for the whole line times (Biel and Glock 2018). In such cases, a

single replacement by naïve operator can cause a significant reduction in the line's

throughput (Biel and Glock 2018; Cohen et al. 2019b; Peltokorpi and Niemi 2019).

The most prevalent mathematical form for learning curve is based on the power

model (Giri and Glock 2017; Peltokorpi and Niemi 2019; Glock et al. 2019) stating that:

𝑡𝑛 = 𝑡1 ⋅ 𝑛−𝑏 (12)

where tn is the execution time of the nth cycle, t1 is the execution time of the first cycle,

and b is the learning constant of the curve. The learning constant of the curve can be

related to the learning slope  (Giri and Glock 2017; Cohen et al. 2019b) by

𝑏 = −𝑙𝑜𝑔2 (𝜙) (13)

25
Thus, using equation (12) it is possible to evaluate the ratio between tn and t1 as:

𝑡𝑛
= 𝑛𝑙𝑜𝑔2 (𝜙) (14)
𝑡1

Also, the total learning time for n repetitions is denoted as Tn, and is computed as:

t1
Tn = n(1−b )
(1 − b) (15)

Moreover, for any task: as n grows tn converges to the standard time of the particular

task tstd. The convergence is rapid, so that after several hundred of repetitions tn

approaches tstd. Bukchin and Cohen (2013) showed that for a given tstd, t1 can be linearly

related to , as shown in Table 5. The implications of Table 5 are important, since it

shows that it is likely that the learning curve would persist for most or all the 8-hour

shift.

Table 5. Relation between the learning slope and the first cycle time
 𝒕𝟏
⁄𝒕 Number of cycles
𝒔𝒕𝒅 to reach tstd
0.7 15 200
0.8 9 900
0.9 3 1300

Since absenteeism is a daily occurrence in long assembly lines, these losses are also

expected to occur daily and consequently a throughput loss of L*THPT is expected.

This loss should be weighed against the number of cobots deployed in the assembly line

(NC). The proposed number of line segments with a cobot. As absence occurs daily in

different workstations of long lines, we assume it happens 250 workdays per year. So,

acquiring and deploying NC cobots in the assembly line is economically justified as

long as:

NC*Lifecost < L* THPT* *250*Life*Uprofit (16)

26
This analysis is illustrated in the following example.

Numerical example:

In this example, three different cases of a typical learning slope in a given line are

computed for:  = {0.7, 0.8, 0.9}. While in all cases the standard time is assumed as one

(1) minute, the corresponding first cycle times from Table 5 are t1 = {15, 9, 3) minutes

respectively. Using equation (13) we have b = {0.512, 0.322, 0.152} respectively, and

using equation (15) for computing the number of cycles n in Tn=480 minutes yields n =

{280, 200, 327}. Assuming this reflects the bottleneck station - n is also the shift

throughput. If all experienced workers are present (no replacement), the throughput (of

8-hour shift) for 1-minute cycles, would simply be n=(8*60)/1=480. Thus, the

corresponding throughput loss during the shift is:

L = {(480-280)/480, (480-200)/480, (480-327)/480} = {42%, 58%, 32%}. In this

example the following values are used to find the limits on the number of cobots for the

assembly line: Lifecost = $ 200,000, Life = 4, THPT = 480, Uprofit = $100

Substituting these values in (16) gives:

200,000*(NC) < L*(480)*250*4*100 → NC < L*240

Thus, for the three values of L={42%, 58%, 32%} the corresponding maximal number

of cobots (with lifetime costs of $200,000 ) in this example is: NC < {100, 139, 76}.

The result is linearly related to all the variables in the right hand-side of (16). For

example, dividing throughput per shift by 10: THPT=480/10=48 and the maximum

number of cobot would change accordingly to NC < {10, 14, 8}.

27
6. Conclusion

This paper describes and discusses the main considerations related to the deployment

decision of a cobot in a manual workstation. It reviews the associated literature of the

related cobots' typical characteristics and capabilities, and discusses their importance.

The type of work the cobot is expected to do, greatly affects the type and cost of the

chosen cobot and must be specified before any productivity or economic analysis. A

productivity based model is developed in sections 4 and 5 for assisting the decision on

acquisition and deployment for a single workstation. A numerical calculation example

follows to illustrate a simple cobot acquisition and deployment justification. However,

in terms of productivity calculations, the case of a single workstation is very different

from the case of a workstation in a sequential assembly line. The need for vastly

different model in sequential assembly line is presented in section 6; followed by an

appropriate model and a computational example. One major conclusion for assembly

lines, is the ability of cobots to alleviate contingencies of absent workers. This "cobot

alleviation" is done in manual stations when a substitute non-experienced worker

replaces an absent worker, and needs to learn the job. The substitute worker is slow

throughout his/her learning stage. The cobot assistance is enabled by prior analysis of

the work, and by preparing contingency pre-programed code, for the cobots to perform

part of the work.

This paper could be directly continued by varied future case studies that discuss

the cobots' contribution to productivity, and improve the productivity-based models

proposed in this paper. In particular it would be interesting to hear about cobots

integration in to I-4.0 systems as part of a whole design of a new workstation. Other

related future research may investigate how to improve cobot deployment by: (1) taking

full advantage of the simultaneous developments in AI, IoT and mobility. (2) enhancing

28
the human cobot interaction, (3) integrating dynamic process planning and coot

operation.

References

Antonelli, D., and Bruno, G. 2019. "Dynamic distribution of assembly tasks in a

collaborative workcell of humans and robots." FME Transactions 47 (4): 723-
730.
Arviv K., Stern H. and Edan Y. 2016. "Collaborative reinforcement learning for a two-
robot job transfer flow-shop scheduling problem", International Journal of
Production Research, 54 (4): 1196-1209.
Ayoubi, Y., Laribi, M. A., Zeghloul, S., and Arsicault, M. 2019. "V2SOM: A novel
safety mechanism dedicated to a cobot’s rotary joints." Robotics 8 (1): 18.
Battaïa, O., Otto, A., Sgarbossa, F., and Pesch, E. 2018. "Future trends in management
and operation of assembly systems: from customized assembly systems to cyber-
physical systems." Omega 78: 1-4.
Bejarano, R., Ferrer, B. R., Mohammed, W. M., and Lastra, J. L. M. 2019. "Implementing
a Human-Robot Collaborative Assembly Workstation." In: proceedings of 17th
International Conference on Industrial Informatics (INDIN) (Vol. 1, pp. 557-
564). IEEE.
Biel, K., & Glock, C. H. 2018. "Governing the dynamics of multi-stage production
systems subject to learning and forgetting effects: A simulation study."
International Journal of Production Research, 56 (10): 3439-3461.
Bogner, K., Pferschy, U., Unterberger, R., and Zeiner, H. 2018. "Optimised scheduling
in human–robot collaboration–a use case in the assembly of printed circuit
boards." International Journal of Production Research 6 (16): 5522-5540.
Bortolini, M., Faccio, M., Gamberi, M., and Pilati, F. 2020. "Motion Analysis System
(MAS) for production and ergonomics assessment in the manufacturing
processes." Computers and Industrial Engineering 139: 105485.
Bortolini, M., Ferrari, E., Gamberi, M., Pilati, F., and Faccio, M. 2017. "Assembly system
design in the Industry 4.0 era: a general framework." IFAC-PapersOnLine 50 (1):
5700-5705.

29
Bragança, S., Costa, E., Castellucci, I., and Arezes, P. M. 2019. "A brief overview of the
use of collaborative robots in industry 4.0: human role and safety." In:
Occupational and Environmental Safety and Health, 641-650. Cham; Springer.
Bukchin, Y., and Cohen, Y. 2013. "Minimising throughput loss in assembly lines due to
absenteeism and turnover via work-sharing." International Journal of Production
Research 51 (20): 6140-6151.
Calzavara, M., Battini, D., Bogataj, D., Sgarbossa, F., & Zennaro, I. 2020. Ageing
workforce management in manufacturing systems: state of the art and future
research agenda. International Journal of Production Research, 58 (3): 729-747.
Cohen, Y., Naseraldin, H., Chaudhuri, A., and Pilati, F. 2019a. "Assembly systems in
Industry 4.0 era: a road map to understand Assembly 4.0." The International
Journal of Advanced Manufacturing Technology 105 (9): 4037-4054.
Cohen, Y., Shoval, S., and Faccio, M. 2019b. "Strategic view on Cobot Deployment in
Assembly 4.0 Systems." IFAC-PapersOnLine. 52 (13): 1519-1524.
de Man, J. C., and Strandhagen, J. O. 2017. "An Industry 4.0 research agenda for
sustainable business models." Procedia CIRP 63: 721-726.
De Winter, J., De Beir, A., El Makrini, I., Van de Perre, G., Nowé, A., and Vanderborght,
B. 2019. "Accelerating Interactive Reinforcement Learning by Human Advice for
an Assembly Task by a Cobot." Robotics 8 (4): 104.
Djuric, A. M., Urbanic, R. J., and Rickli, J. L. 2016. "A framework for collaborative robot
(CoBot) integration in advanced manufacturing systems." SAE International
Journal of Materials and Manufacturing 9 (2): 457-464.
El Makrini, I., Elprama, S. A., Van den Bergh, J., Vanderborght, B., Knevels, A. J.,
Jewell, C. I., and Berte, J. 2018. "Working with walt: How a cobot was developed
and inserted on an auto assembly line." IEEE Robotics & Automation Magazine
25 (2): 51-58.
El Zaatari, S., Marei, M., Li, W., and Usman, Z. 2019. "Cobot programming for
collaborative industrial tasks: An overview." Robotics and Autonomous Systems
116: 162-180.
Faccio, M., Bottin, M., and Rosati, G. 2019. "Collaborative and traditional robotic
assembly: a comparison model." The International Journal of Advanced
Manufacturing Technology, 102 (5-8): 1355-1372.

30
Faccio, M., Minto, R., Rosati, G., & Bottin, M. (2020). "The influence of the product
characteristics on human-robot collaboration: a model for the performance of
collaborative robotic assembly." The International Journal of Advanced
Manufacturing Technology 106 (5): 2317-2331.
Fager, P., Calzavara, M., and Sgarbossa, F. 2020. "Modelling time efficiency of cobot-
supported kit preparation." The International Journal of Advanced Manufacturing
Technology 106 (5): 2227-2241.
Fast-Berglund, Å., Palmkvist, F., Nyqvist, P., Ekered, S., and Åkerman, M. 2016.
"Evaluating Cobots for Final Assembly." Procedia CIRP 44: 175-180.
Fiaidhi, J., Mohammed, S., and Mohammed, S. 2018. "The robotization of extreme
automation: The balance between fear and courage." IT Professional 20 (6): 87-
93.
Frank, A. G., Dalenogare, L. S., and Ayala, N. F. 2019. "Industry 4.0 technologies:
Implementation patterns in manufacturing companies." International Journal of
Production Productivity 210: 15-26.
Frick, B., Simmons, R., and Stein, F. 2018. "The cost of shift work: Absenteeism in a
large German automobile plant." German Journal of Human Resource
Management 32 (3-4): 236-256.
Gil-Vilda, F., Sune, A., Yagüe-Fabra, J. A., Crespo, C., and Serrano, H. 2017.
"Integration of a collaborative robot in a U-shaped production line: a real case
study." Procedia Manufacturing 13: 109-115.
Giri, B. C., & Glock, C. H. 2017. "A closed-loop supply chain with stochastic product
returns and worker experience under learning and forgetting." International
Journal of Production Research 55 (22): 6760-6778.
Glock, C. H., Grosse, E. H., Jaber, M. Y., and Smunt, T. L. 2019. "Applications of
learning curves in production and operations management: A systematic literature
review." Computers & Industrial Engineering 131: 422-441.
Grosse, E. H., Glock, C. H., Jaber, M. Y., and Neumann, W. P. 2015. "Incorporating
human factors in order picking planning models: framework and research
opportunities." International Journal of Production Research 53 (3): 695-717.
Guo, D., Zhong, R. Y., Ling, S., Rong, Y., and Huang, G. Q. 2020. "A roadmap for
Assembly 4.0: self-configuration of fixed-position assembly islands under

31
 Graduation Intelligent Manufacturing System," International Journal of
Production Research 58 (15): 4631-4646.
Gustavsson, P., Holm, M., Syberfeldt, A., and Wang, L. 2018. "Human-robot
collaboration–towards new metrics for selection of communication
technologies." Procedia CIRP 72: 123-128.
Hashemi-Petroodi, S. E., Dolgui, A., Kovalev, S., Kovalyov, M. Y., & Thevenin, S.
2020. "Workforce reconfiguration strategies in manufacturing systems: a state of
the art." International Journal of Production Research: 1-24.
doi:10.1080/00207543.2020.1823028
Heo, Y. J., Kim, D., Lee, W., Kim, H., Park, J., and Chung, W. K. 2019. "Collision
detection for industrial collaborative robots: a deep learning approach." IEEE
Robotics and Automation Letters 4 (2): 740-746.
Hoffman, G. 2019. "Evaluating fluency in human–robot collaboration." IEEE
Transactions on Human-Machine Systems, 49 (3): 209-218.
Ivanov, D., Tang, C. S., Dolgui, A., Battini, D., & Das, A. 2020. "Researchers'
perspectives on Industry 4.0: multi-disciplinary analysis and opportunities for
operations management." International Journal of Production Research: 1-24.
doi:10.1080/00207543.2020.1798035
Jiao, J.R., Zhou, F., Gebraeel N.Z. and Duffy V.2020. "Towards augmenting cyber-
physical-human collaborative cognition for human-automation interaction in
complex manufacturing and operational environments." International Journal of
Production Research 58 (16): 5089-5111.
Kadir, B. A., Broberg, O., and Souza da Conceição, C. 2018. "Designing human-robot
collaborations in industry 4.0: explorative case studies." In: DS 92: Proceedings
of the DESIGN 2018 15th International Design Conference: 601-610. doi:
/[Link]/10.21278/idc.2018.0319
Kildal, J., Tellaeche, A., Fernández, I., and Maurtua, I. 2018. "Potential users' key
concerns and expectations for the adoption of cobots." Procedia CIRP 72: 21-26.
Liu, H., and Wang, L. 2018. "Gesture recognition for human-robot collaboration: A
review." International Journal of Industrial Ergonomics 68: 355-367.
Liu, Q., Liu, Z., Xu, W., Tang, Q., Zhou, Z., and Pham, D. T. 2019. "Human-robot
collaboration in disassembly for sustainable manufacturing." International
Journal of Production Research 57 (12): 4027-4044.

32
Malik, A. A., & Bilberg, A. 2017. "Framework to implement collaborative robots in
manual assembly: a lean automation approach." In B. Katalinic (Ed.),
Proceedings of the 28th DAAAM International Symposium, Published by
DAAAM International, January: 1151-1160.
Malik, A. A., & Bilberg, A. 2019. "Complexity-based task allocation in human-robot
collaborative assembly." Industrial Robot: the international journal of robotics
research and application 46 (4): 471–480
Markets And Markets, 2019. "Collaborative Robot Market, Market Research Report: Nov
2019," Report Code: SE 4480, [Link]
Reports/[Link], (retrieved Nov. 2020).
Marvel, J. A., and Norcross, R. 2017. "Implementing speed and separation monitoring in
collaborative robot workcells." Robotics and computer-integrated manufacturing
44: 144-155.
Matheson, E., Minto, R., Zampieri, E. G., Faccio, M., and Rosati, G. 2019. "Human–
Robot Collaboration in Manufacturing Applications: A Review." Robotics, 8 (4):
100.
McClure, P. K. 2018. "“You’re Fired,” Says the Robot: The Rise of Automation in the
Workplace, Technophobes, and Fears of Unemployment." Social Science
Computer Review 36 (2): 139-156.
Mecheri, S. S., and Greene, C. M. 2019. "Collaborative robot selection using analytical
hierarchical process." International Journal of Rapid Manufacturing 8 (4): 326-
344.
Michaelis, J. E., Siebert-Evenstone, A., Shaffer, D. W., and Mutlu, B. 2020, April.
"Collaborative or Simply Uncaged? Understanding Human-Cobot Interactions in
Automation.". In: Proceedings of the 2020 CHI Conference on Human Factors in
Computing Systems, Honolulu, April 1-12. CHI.
Moeuf, A., Lamouri, S., Pellerin, R., Tamayo-Giraldo, S., Tobon-Valencia, E., and
Eburdy, R. 2020. "Identification of critical success factors, risks and opportunities
of Industry 4.0 in SMEs." International Journal of Production Research 58(5):
1384-1400.
Ostanin, M., Yagfarov, R., Devitt, D., Akhmetzyanov, A., and Klimchik, A. 2020. "Multi
robots interactive control using mixed reality." International Journal of
Production Research, Ahead of print: 1-13.

33
 doi:10.1080/00207543.2020.1834640
Pearce, M., Mutlu, B., Shah, J., and Radwin, R. 2018. "Optimizing makespan and
ergonomics in integrating collaborative robots into manufacturing processes."
IEEE Transactions on Automation Science and Engineering, 15 (4): 1772-1784.
Peltokorpi, J., and Niemi, E. 2019. "Effects of group size and learning on manual
assembly performance: an experimental study." International Journal of
Production Research 57 (2): 452-469.
Pollak, A., Paliga, M., Pulopulos, M. M., Kozusznik, B., and Kozusznik, M. W. 2020.
"Stress in manual and autonomous modes of collaboration with a cobot."
Computers in Human Behavior, 106469.
Romero, D., Stahre, J., Wuest, T., Noran, O., Bernus, P., Fast-Berglund, Å., & Gorecky,
D. 2016.). "Towards an operator 4.0 typology: a human-centric perspective on the
fourth industrial revolution technologies." In proceedings of the international
conference on computers and industrial engineering (CIE46), Tianjin, China,
October: 29-31.
Sadik, A. R., and Urban, B. 2017. "An ontology-based approach to enable knowledge
representation and reasoning in worker–cobot agile manufacturing." Future
Internet 9 (4): 90-113.
Seriani, S., Gallina, P., Scalera, L., and Lughi, V. 2018. "Development of n-DoF
preloaded structures for impact mitigation in cobots." Journal of Mechanisms and
Robotics, 10 (5): 1-12.
Sherwani, F., Asad, M. M., and Ibrahim, B. S. K. K. 2020. "Collaborative Robots and
Industrial Revolution 4.0 (IR 4.0). In: 2020 International Conference on
Emerging Trends in Smart Technologies (ICETST): 1-5. IEEE.
Simões, A. C., Soares, A. L., and Barros, A. C. 2020. "Factors influencing the intention
of managers to adopt collaborative robots (cobots) in manufacturing
organizations." Journal of Engineering and Technology Management 57, 101574.
Tamas, L., and Murar, M. 2018. "Smart CPS: vertical integration overview and user story
with a cobot." International Journal of Computer Integrated Manufacturing: 1-
18.
Vicentini, F. 2020. "Terminology in safety of collaborative robotics." Robotics and
Computer-Integrated Manufacturing 63, 101921.

34
 Vojić, S. 2020. "Applications of collaborative industrial robots." Machines,
Technologies, Materials 14 (3): 96-99.
Zhong, H., Wachs, J. P., and Nof, S. Y. 2013. "A collaborative telerobotics network
framework with hand gesture interface and conflict prevention." International
Journal of Production Research 51 (15): 4443-4463.

35

View publication stats