Available online at www.sciencedirect.

com

ScienceDirect
Procedia CIRP 106 (2022) 96–101
www.elsevier.com/locate/procedia

9th CIRP Conference on Assembly Technology and Systems

Innovative robotization of manual manufacturing processes

Peter Chemwenoa*, Robbert-Jan Torna
a
Chair of advanced manufacturing, Faculty of Engineering Technology, University of Twente, Enschede 7522 NB, The Netherlands,
* Corresponding author. E-mail address: [email protected]

Abstract

Manufacturing industries are continually challenged to adapt to a competitive environment. Consequently, there is an urgency to opt for
automation technologies to upgrade their manufacturing facilities and make them more flexible. Especially this relates to automating manual
manufacturing processes, which is often challenging to structure to ensure repetitiveness and generalization to other processes within the facility.
Consequently, innovating systematic approaches to identify robotization opportunities is an interesting proposition for manufacturing set-ups that
would like to integrate collaborative robots on the shop floor and struggle with the decision steps to follow.

In this paper, a framework for supporting robotization effort for manufacturing set-ups is proposed. The methodology consisting of five phases,
culminating in the identification of robotization opportunities. A case for manual milling manufacturing processes is demonstrated as a ‘proof-
of-concept’. The first step of the proposed approach focuses on task decomposition, in which manual manufacturing tasks are characterized. This
is followed by task allocation to a robot and human agent based on intrinsic characteristics of the task to capabilities of the agent. Next, alternative
layout configurations for candidate cell layouts are generated. In the final step, a candidate layout is selected and modeled in an agent-based
simulation platform, considering factors such as realism, interaction safety between the robot and human agent, and interesting manufacturing
metrics such as resource utilization and throughput rate. A final configuration is optimized, which visualizes a collaborative robot performs
loading and unloading tasks alongside an operator performing highly cognitive tasks. For safety, zoning of the manufacturing cell is visualized,
considering a working area separated by a safety fence.
© 2022 The Authors. Published by Elsevier Ltd.
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientific committee of the 9 th CIRP Conference on Assembly Technology and Systems

Keywords: Robotisation; manual manufacturing tasks; structured decision making; manufacturing cell layout; agent-based simulation; safety

1. Introduction forward to solutions yielding maximum benefits while

investing the least effort, time, and money. The trend of
The rapidly changing market conditions are posing concepts like Industry 4.0, Cyber-Physical System (CPS),
challenges to the manufacturing industries to remain Internet of Things (IoT), Robotization, and Robotic Process
competitive in high-wage locations. The key factors that will Automation (RPA) shows the increasing reliance on industrial
help the companies survive in this globalized market are high- automation solutions to sustain competitive advantage among
quality products, on-time deliveries, maintaining low the manufacturing industries. Industrial automation holds the
production costs, and producing customized products [1]. potential to assist the manufacturing industries in achieving
Improving the manufacturing system’s flexibility is believed to sustainable competitive advantage [2].
play a substantial role in attaining the factors above. Emerging Industrial automation focuses on automating production
technologies available in the manufacturing domain give the processes and systems by substituting human workers with
manufacturing companies ample choices to invest and gain mechanical, electronic, computer-based systems, robots, and
benefits. Nowadays, most manufacturing companies look information systems to operate and to monitor manufacturing

2212-8271 © 2022 The Authors. Published by Elsevier Ltd.

This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientific committee of the 9 th CIRP Conference on Assembly Technology and Systems
10.1016/j.procir.2022.02.161

This is a resupply of March 2023 as the template used in the publication of the original article contained errors. The content of the article has remained unaffected.
 Peter Chemweno et al. / Procedia CIRP 106 (2022) 96–101 97

[3]. The computer-based systems, information, and electronic transform manual manufacturing processes into a fully
systems automate cognitive tasks, whereas the robots and robotized, or semi-robotized solution? To answer this question,
mechanical systems automate the physical task in we propose a robust decision-making framework to assist
manufacturing systems. International Federation of Robotics manufacturing facilities in identifying the potential of
(IFR) reports a worldwide increase of 85% in industrial robots transforming the manual manufacturing process into a semi-
worldwide within five years (2014-2019), thus a total of 2.7 robotized solution. We present a proof of concept whereby
million industrial robots operating globally in 2020 [4]. identified solutions can be visualized prior to
Apart from providing economic benefits, the robot can designing/implementing an actual robotized manufacturing
perform tasks that require handling heavy objects, high cell. Furthermore, the proposed framework allows decision-
precision, high quality, operating in dangerous conditions, and makers to evaluate the feasibility of robotization, from an
improving ergonomics. The innovations in the field of robotics implementation perspective.
broaden the scope of the application of robots in the Fig. 1 illustrates the framework we propose for robotizing
manufacturing industry. manual manufacturing processes. The framework contains four
We often perceive the decision of integrating robots in the sequential functions discussed in Sections 2 and 3. The arrows
manufacturing system as either transformation into a fully in the framework represent data and actors interrelating with the
automated or semi-robotized system or keeping it entirely functions. The horizontal arrows represent the in- and output of
manual [5]. While the robots excel in delivering high-quality data, whereas the vertical arrows indicate restrictions (could be
work repeatedly, it does not provide high flexibility when the subjective to change) and knowledge provided by actors.
production demands change quickly. In contrast, Humans offer
the abilities of adaptability, dexterity, and in-process decision- 2. Current state of research
making skills. A manufacturing system can be a manual, fully
automated, or Hybrid system (i.e., a combination of manual and 2.1. Robotization of manufacturing
automated systems). When the utilization of robots and humans
is according to the process requirement, the manufacturing “Robotization” is the terminology used when a robot is
system attains the highest efficiency [6]. employed to automate manual tasks. The manufacturing
Thus, integrating robots with a manufacturing system in the industry witnessed an increase in industrial robot adaption into
absence of a structured framework may be fraught with risks, its production processes due to its capability to perform tasks
especially with overly relying on tacit knowledge of decision- with high precision and repeatability. The ability of the
makers at manufacturing facilities. Additional open questions industrial robot to work continuously helped manufacturers to
include verifying prior to implementation, and the feasibility of increase output. Robots can work in dangerous and harmful
a robotization solution considering safety aspects. conditions, thus improving the manufacturing system’s
This paper addresses some of these open questions and, more working environment and safety. Thus, industrial robots’
specifically, the research question: How can decision-makers various advantages encourage the manufacturing industry to

Fig. 1. Framework for robotizing manual manufacturing

This is a resupply of March 2023 as the template used in the publication of the original article contained errors. The content of the article has remained unaffected.
 98 Peter Chemweno et al. / Procedia CIRP 106 (2022) 96–101

incorporate industrial robots into their manufacturing systems 3. Methodology

to improve their productivity and profitability.
Robotization in manufacturing sub-divides into two parts 1) In this study, the methodology depicted in Fig. 1 is applied.
robotization of processing operations and 2) robotization of The first main step evaluates challenges and opportunities for
handling operations. Handling is the major robot application robotization, where expert input is important. For the case
field found in all the areas of manufacturing [7]. The organization, suggestions on low hanging robotization
robotization of processing operations considers using an opportunities were evaluated via brain storms, prior to agreeing
industrial robot to carry out tasks such as welding, painting, on labor intense, but low cognition tasks.
machining, inspection, and assembly operations. The handling The next sub-steps consists of task decomposition, analysis,
operations such as pick and place, machine tending, packing, allocation and designing a virtual workspace integrating
palletizing, and transporting utilizes an industrial robot in the industrial robots. The sub-steps are discussed further below.
robotization of handling operations.
The introduction of collaborative robots helped 3.1. Task decomposition, analysis, and allocation
manufacturing industries to overcome the limitation mentioned
above. A collaborative robot, also known as “Cobots,” is a For task decomposition, the Hierarchical Task Analysis
robotic device that collaborates with humans. Collaborative (HTA) is a widely used method for structuring manufacturing
robots share the workspace and work alongside humans tasks, and allowing decision-makers to extensive characterize
without the requirement of safety fences. The combination of manufacturing tasks. It does this by decomposing high-level
collaborative robot and human allows taking advantage of the tasks into lower-level sub-tasks up to basic task elements [13].
robot’s ability to work with high accuracy with speed and HTA method breaks the tasks into simplified sub-tasks.
repetitiveness, along with flexibility and cognitive skills of the According to Kirwan and Ainsworth [14], HTA is the best-
human [8]. known task analysis technique to understand and analyze a
All the collaborative robots have an inbuilt safety system in manufacturing system.
the form of sensors and foam, which guarantees that the cobots In this study, existing manual steps were decomposed, from
will stop and safeguard humans when a collision occurs. Four higher-level to lower-level sub-tasks, thus simplifying complex
types of collaborative robots are mentioned in ISO/TS 15066 human-relevant task steps/actions into a sequence of simplified
based on human interaction with the cobots [9, 10]. This task steps. To support HTA, different data capture techniques
includes: are used and, most prominently, video recording (and analysis)
of manufacturing processes or detailed process mapping.
• Safety Monitored Stop: Includes measures to a human’s
For this study, we assume that manufacturing tasks have a
presence in a collaborative workspace. hierarchical relationship, which often may not the case,
• Hand Guiding: Hand guiding robot’s motion is only especially for customized products where task steps are
possible using direct input of the operator. dynamic. However, since this project relates to a standardized
• Speed and Separation Monitoring: Influences the robot product, whose tasks are static, we view the HTA as rather
motion and adapts the manipulator speed when an operator robust.
enters the shared workspace.
• Power and force limitation: Limits the forces exerted by 3.2. Task allocation
the robot manipulator to a level below thresholds that would
be harmful. Once decomposed, tasks were allocated to either the human
or robotic agent depending on the capabilities of the specific
2.2. Challenges and opportunities robotizing manufacturing agent [15]. As an example, high cognition task elements are
best suited for the human operator, with repetitive tasks
A robotic manufacturing cell presents an attractive solution as allocated to a robot agent. For allocation, criteria were
many repetitive and manual tasks are candidates for considered, such as payload (weight), cognitive demand of the
robotization. However, it is crucial to consider the criteria used task, cycle time, ergonomics, and investment requirements for
to robotize tasks: task characteristics, robot characteristics, and actuators needed to substitute performance of the task step.
robot capabilities. Task characteristics give information about Additional allocation criteria applied in the study, include
the task, such as the task’s nature (i.e., Physical or cognitive), product size, shape and weight ; cycle time, repetitiveness, task
task time, and task complexity. The information about robot complexity; payload, repeatability, reachability and
characteristics and capabilities such as accuracy, payload, ergonomics; cognitive and physical strength [16, 17].
reachability, degrees of freedom, and work envelope help
check the robot’s suitability for performing a task [11]. 3.3. Designed robotized manufacturing cell
As a first step for identifying robotization opportunities,
Shepard emphasized the importance of task analysis as a means After task allocation, designing the work cell is the next
of understanding which task elements are suitable to be feasible step. This involves designing a cell layout considering
performed by a human operator or a collaborative robot [12]. the positioning of the agents and trading off with system
More importantly, the analysis allows decision-makers to performance. The idea is to design a layout that optimizes
characterize manufacturing processes based on the capabilities workflow within the cell. For this study, an Agent-based
of an agent, here the human operator or the collaborative robot. simulation (ABS) was implemented, as it presents an attractive
approach for designing and visualizing robotized

This is a resupply of March 2023 as the template used in the publication of the original article contained errors. The content of the article has remained unaffected.
 Peter Chemweno et al. / Procedia CIRP 106 (2022) 96–101 99

manufacturing cell layouts [18]. It allows the modeler to • High-level task 1: preparing the press-brake machine tool
represent the cell configuration realistically and experiment for loading
with alternative layouts before defining a more optimal • Sub-task 1.1: Cleaning the press-brake tool
configuration. Moreover, the optimization can consider • Sub-task 1.2: Changing the fixture for the holding tool
manufacturing metrics and safety zoning to prevent
unanticipated collisions with the robotic agent [19]. Overall, the press brake tool milling was decomposed into five
high-level tasks and a total of 148 sub-ordinate tasks.
4. Implementation of use cases of milling processes
4.2. Task criteria, rationalization, and allocation
The proposed methodology was implemented for robotizing
manual milling processes for press brake tooling for bending Prior to allocating the sub-ordinate tasks to the specific agent
sheet metal. It consists of five (5) steps: (human or robot) based on capabilities, several allocation
criteria were defined based on expert intuition and augmented
4.1. Task analysis by a literature search. The first criteria considered product
characteristics, including size, shape, geometry, and weight,
Task analysis is implemented to characterize manual milling which influenced actuation aspects such as grasp-ability,
task steps. The Hierarchical Task Analysis (HTA) method was maximum allowable gripper payload, and manipulation ease of
applied to decompose manual tasks performed by the milling the product by a gripper. Fig. 3 illustrates the tool geometry,
cell operator in a well-defined and sequential order. For which influences the ‘graspability’ criterion.
collecting information, video recordings of the process were The second criteria considered characteristics of the
captured, and afterward, tasks were decomposed and tabulated. production task. Cycle time, repetitiveness, and required
precision were identified as important. It is important to note
that the criteria were either quantitative or qualitative,
influencing how each criterion was allocated to the sub-
ordinate tasks. Table 2 illustrates an example of task allocation
following a ‘quick-and-dirty’ qualitative assessment of the
appropriateness of a criterion to the capabilities of an
implementation agent.

Table 2. Illustration of hierarchical decomposition for a high-level manual

milling task

Fig. 2. (a) cleaning the press-brake tool; (b) changing fixture for holding tool.
Sub-ordinate Grasp- Need to
Shape Repetitive? Agent?
tasks ability? manipulate?
Fig. 2 illustrates examples of decomposed manual task steps
based on video recordings of the entire milling process for press
1.1.1. Cuboid Yes Yes No Robot
brake tooling. A hierarchical representation of the two sub-
tasks is illustrated in Table 1, with Task 1 “preparing the press 1.1.2. Cuboid Yes Yes No Robot
brake tool for loading on the milling machine” being the super- 1.1.4. Cylinder Yes Yes No Operator
ordinate tasks, while sub-tasks (a) and (b) illustrated in Fig. 2 Cylinder Yes Yes No Operator
1.1.5.
representing the decomposed sub-tasks, and sub-ordinate tasks.

Table 1. Illustration of hierarchical decomposition for a high-level manual Moreover, a task ‘rationalization is implemented to standardize
milling task decomposed sub-ordinate tasks, and ensure tasks are performed
High- irrespective of product variety. Examples include standardizing
Sub-
Level
tasks
Sub-ordinate tasks the tool holding fixture, which influences the loading process
Task of tooling by the operator on the fixture.
1.1.1 Hold the press brake tool with two hands
1.1.2 Rotate the press brake tool to a vertical position
1.1.3 Hold the press brake tool vertically
1.1
1.1.4 Pick up the bottle of cleaner liquid from working table
1.1.5 Position cleaner on cleaning fabric
1. 1.1.6 Spray the cleaner liquid on the fabric
1.2.1 Pick up the screwing drill
1.2.2 Move it to the screw on the lower clamp
1.2 1.2.3 Place it on the head of the screw
1.2.4 Press the power button
Fig. 3. Press brake tooling geometry.
1.2.5 Once unscrewing is done, stop the drill

This is a resupply of March 2023 as the template used in the publication of the original article contained errors. The content of the article has remained unaffected.
 100 Peter Chemweno et al. / Procedia CIRP 106 (2022) 96–101

4.3. Designing feasible conceptual cell configurations throughput and resource utilization. Other considerations were
space allocation while minimizing changes to the existing
After tasks were rationalized and allocated to the operator and manufacturing facility.
collaborative robot, feasible cell configurations were
generated, consideration capabilities and operator safety.
generated and evaluated further on space, resource utilization,
and operator access [20]. Fig. 5 illustrates one of the conceptual
layouts considering the positioning of a robot manipulator
picking and position press brake tooling on two milling
machines. The operator prepares the tooling on a workbench
and places it on a pallet. The robot arm subsequently picks the
tooling from the pallet and position it on an automated pallet
changer (APC), which feeds the tooling to the milling machine.
After milling, the robot arm picks and places the tooling on the
outbound APC, whereafter the operator picks and stores it on a
storage system.
Following the design thinking approach, five concepts were
Agent-based simulation of cell configuration
In the next step, the five alternative concepts were visualized in
Visual Components, an agent-based simulation software [21]. Fig. 4. Real-time statistics of operator utilization (red circle indicating breaks)
Input considerations for the model included parameters such as
positioning and distance of the resources, task cycle times,
5. Discussion
operator walking speed, machine processing times, and
operator schedules. Input values were based on empirical
In this study, we propose a structured framework for
process times at the use case organization.
identifying robotization opportunities for manual
Several performance measures were defined to consider
manufacturing processes. This study addresses an automation
system performance to compare the modeled cell
challenge: how can decision-makers identify robotization
configurations. This includes resource/agent utilization, and
possibilities for manual processes in a structured and time-
production throughput. Fig. 5 shows the simulation model of
efficient way? Furthermore, a question addressed relates to the
one of the alternative layouts. Fig. 4 illustrates real-time
implementation of feasible concepts, where in this study, an
statistics of the operator utilization for one of the
agent-based simulation approach is proposed. The proposed
configurations. Based on the evaluation of the performance
approach can be applied at both the starting phase of a new
measures, an optimal configuration was selected considering
automation challenge or translating existing processes into
objective functions, including maximizing production
semi or fully automated robotized solutions.

Fig. 5. Illustration of simulated cell configuration.

This is a resupply of March 2023 as the template used in the publication of the original article contained errors. The content of the article has remained unaffected.
 Peter Chemweno et al. / Procedia CIRP 106 (2022) 96–101 101

As seen in the study, to identify robotization opportunities, [8] Villani V, Pini F, Leali F, Secchi C. Survey on human–robot collaboration
it is essential to understand the manufacturing tasks currently in industrial settings: Safety, intuitive interfaces and applications.
Mechatronics. 2018 Nov 1;55:248-66.
performed manually to identify the ideal task steps for [9] ISO/TS 15066:2016 Robots and robotic devices- Collaborative robots
robotization. This understanding is important for task 2016.
allocation decisions (i.e., physical or cognitive tasks), thus [10] Chemweno, P., L. Pintelon, and W.J.S.S. Decre, Orienting safety
forming the basis for allocating the task to the robot or the assurance with outcomes of hazard analysis and risk assessment: A review
operator. The Hierarchical Task Analysis (HTA) approach is of the ISO 15066 standard for collaborative robot systems. 2020. 129: p.
104832.
particularly useful to decompose the high-level tasks into sub- [11] Torn, R.-J., Chemweno, P., Vaneker, Soheil., Arastehfar, Soheil, Towards
tasks and subordinate tasks structurally. a structured decision-making framework for automating cognitively
Furthermore, agent-based simulation is useful at the demanding manufacturing tasks in CARV, 2021, Towards Sustainable
implementation phase, especially for modeling and visualizing Customization: Bridging Smart Products and Manufacturing Systems.
alternative layouts of new semi-robotized solutions. The 2021: Aalborg City University, Denmark.
[12] Shepard, A., Task Analysis, in Evaluation of Human Work. 2015, Taylor
simulation also makes it possible to compare alternative & Francis Group. p. 139-162.
configurations using performance indicators, including CNC [13] Jenkins, D.P., et al., Human Factors Methods A Practical Guide for
machine’s utilization, robot’s utilization, operator’s utilization, Engineering and Design. 2013: Ashgate.
and the throughput for the alternative configurations. This [14] Kirwan, B. and L.K.Ainsworth, A Guide to Task Analysis. 1992, London:
provides valuable insights prior to actual implementation. Taylor & Francis.
[15] Cheng Y, Sun F, Zhang Y, Tao F. Task allocation in manufacturing: a
review. Journal of Industrial Information Integration. 2019 Sep 1;15:207-
18.
6. Conclusions [16] Blankemeyer S, Recker T, Stuke T, Brokmann J, Geese M, Reiniger M,
Pischke D, Oubari A, Raatz A. A method to distinguish potential
For future work, the study will define more dynamic and workplaces for human-robot collaboration. Procedia CIRP. 2018 Jan
1;76:171-6.
quantifiable criteria for allocating tasks to agents prior to [17] Neb, A. and D. Remling, Quantification and Evaluation of Automation
designing alternative layout configurations. This would yield a Concepts based on a Multi-Criteria Decision Analysis. 29th CIRP Design
robust allocation criterion, and better insights on task 2019 (CIRP Design 2019), 2019. 84: p. 624-629.
allocation. Additional work will focus on defining task analysis [18] Malik, A.A. and A. Bilberg. A framework to implement collaborative
rules, and especially a ‘stop rule for complex manufacturing robots in manual assembly: a lean automation approach. in 28th daaam.
2017. vienna, austria.
tasks. Often, the hierarchical task levels can be extensive and [19] Tsarouchi P, Michalos G, Makris S, Athanasatos T, Dimoulas K,
may complicate the task allocation steps. Future steps to Chryssolouris G. On a human–robot workplace design and task allocation
improve the modeling steps for the agent-based simulation will system. International Journal of Computer Integrated Manufacturing. 2017
focus on developing dynamic models, considering stochasticity Dec 2;30(12):1272-9.
in manufacturing processes. [20] Brenner, W., F. Uebernickel, and T. Abrell, Design thinking as mindset,
process, and toolbox, in Design thinking for innovation. 2016, Springer. p.
3-21.[21] Components, V. Design of Factories of the Future. 2021 25th
7. Acknowledgements July, 2021]; Available from: https://www.visualcomponents.com/.

We would like to acknowledge Jay Chauhan, Faculty of

Engineering Technology, University of Twente, for his
contribution to this study, setting up the experiment, and
implementing the safety zoning approach. We also thank Frank
Rouweler for his contribution as well.

References

[1] Müller, R., M. Vette, and O. Mailahn. Process-oriented task assignment for
assembly processes with human-robot. in Conference on Assembly
Technologies and Systems (CATS). 2016. Elsevier.
[2] Acharya, V., S.K. Sharma, and S.K. Gupta, Analyzing the factors in
industrial automation using analytic hierarchy process. Computers &
Electrical Engineering (VOL:7), 2018: p. 877-886.
[3] Säfsten, K., M. Winroth, and J. Stahre, The content and process of
automation strategies. Int. J. Production Economics, 2007: p. 25–38.
[4] Tantawi KH, Sokolov A, Tantawi O. Advances in industrial robotics: From
industry 3.0 automation to industry 4.0 collaboration. In2019 4th
Technology Innovation Management and Engineering Science
International Conference (TIMES-iCON) 2019 Dec 11 (pp. 1-4). IEEE.
[5] Winroth, M. and K. Säfsten. Automation strategies - implications on
strategy process from refinement of manufacturing strategy content. in
POMS 19th annual conference. 2008. California, U.S.A.
[6] Frohm, J., V. Lindström, and M. Bellgran. A model for parallel levels of
automation within manufacturing. in International Conference on
Production Research. 2005.
[7] Wilson, M., Typical Applications, in Implementation of Robot Systems.
2015, Butterworth-Heinemann. p. 75-102.

This is a resupply of March 2023 as the template used in the publication of the original article contained errors. The content of the article has remained unaffected.