Manufacturing Rev.

11, 17 (2024)
© J. Xiao and K. Huang, Published by EDP Sciences 2024
https://doi.org/10.1051/mfreview/2024015
Available online at:
https://mfr.edp-open.org

REVIEW

A comprehensive review on human–robot collaboration

remanufacturing towards uncertain and dynamic disassembly
Jinhua Xiao* and Kaile Huang
School of Transportation and Logistics Engineering, Wuhan University of Technology, Wuhan 430063, PR China

Received: 30 December 2023 / Accepted: 3 June 2024

Abstract. With the continuously increasing volumes of End-of-Life (EOL) products and the emergence of
Industry 4.0-related technologies, the manual labor-intensive disassembly in remanufacturing process is
gradually shifting towards human–robot collaboration (HRC) disassembly. However, it is necessary to consider
the most commonly approach in current robot-involved automatic disassembly with the high efficiency and
adaptability. The purpose of this paper is to summarize the existing human–robot collaboration disassembly
technologies to further discuss the possible applications of various technologies in the disassembly process,
thereby providing the comprehensive reviews of future research directions. The paper presents an analysis of the
technical issues encountered in the HRC disassembly process, which provides a summary of the intelligent
disassembly platform with robot agents as the core, focusing on the optimization of HRC disassembly.
Furthermore, the relevant technologies are reviewed to discuss the intelligent disassembly systems, including
Cyber-Physical systems (CPS), Internet of Things (IoT), Cloud Computing, Artificial Intelligence (AI), and
Digital Twins (DT). The industrial applications of various optimization methods have discussed the potential
research trends in the collaboration of multiple robots and humans in disassembly with the perspectives of task
allocation, collaborative decision-making, and performance evaluation, focusing on the application trends of
digital twin technology in industrial disassembly lines.
Keywords: Human–robot collaboration / disassembly / industry 4.0 / digital twin / uncertainty

1 Introduction However, remanufacturing process typically includes

six stages: product recovery, initial inspection, disassem-
With the increasing global development of sustainability bly, component repair, reassembly, and performance
and environmental protection, industrial green testing [3]. Disassembly, as one of the most key steps in
manufacturing has become increasingly urgent. Simulta- the remanufacturing system, holds significant value to
neously, the demand for green products by consumers has accomplish the green manufacturing in the recycling. With
been rising, which facing with the recycling of End-of-Life the increasing volumes of EOL products, there is a need to
(EOL) products to subsequently develop the gradual achieve full or partial automation of disassembly process
emergence of the remanufacturing industry [1]. Remanu- to improve efficiency in the remanufacturing process [4].
facturing involves the repair, refurbishment, or disassem- Due to the uncertainty associated with EOL products and
bly/reassembly of products to restore their original the requirements of completely automation production,
functionality or meet new product requirements, thereby the implementation of fully automation disassembly lines
extending product lifespans to reduce environmental remains many challenging [5,6]. Similarly, the semi-
impacts. Remanufacturing products can be priced up to automated disassembly lines offer new approaches to
50% lower than new products [2] to provides the consumers improve disassembly efficiency according to their low
with more attractive options. However, it is necessary to investment costs, strong flexibility, and good adaptability
not only protect the environment to achieve efficient [7]. The typical semi-automated disassembly mainly
resource utilization, but also provide strong impetus for focuses on human–robot collaborative (HRC) disassembly,
economic and social sustainability. where human operators and robots agents jointly in the
disassembly operations to achieve accurate disassembly of
EOL products. Similarly, the HRC modes in disassembly
* e-mail: [email protected] can be divided into three types as follows:

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0),
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
2 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024)

– Sequential disassembly: as a process of gradually automation and intelligent technologies has enhanced the
disassembling products in a specific order, typically used flexibility and adaptability of the industrial robots in
for more structured and orderly disassembly tasks. remanufacturing process, thus achieving the closer HRC
– Parallel disassembly: as simultaneous disassembly of disassembly. With the recent advancements of the
multiple components, involving different personnel or intelligent collaborative technologies, remanufacturing
robots to enhance disassembly efficiency. process can be used to improve the potential values of
– Collaborative disassembly: as a mode where humans and EOL product. However, the diversity of EOL products has
robots cooperate in disassembly, leveraging each other’s brought out the impact of the uncertainty, leading to
strengths to enhance overall disassembly efficiency and increased complexity of HRC disassembly, while also
safety. These three forms of disassembly each have their affecting overall time and cost. Therefore, it is essential to
own advantages to different disassembly scenarios. summarize the scientific challenges of HRC disassembly
technologies.
By comparing the traditional disassembly processes,
the new disassembly methods based on HRC operations are
faced with the higher safety risks and the stricter recycling 2.1 Uncertain product quality
requirements [8]. Collaborative work by human operators
and robot agents in the same working space may generate Almost all the EOL products undergo disassembly to
the risks of collision and entanglement, especially in the accomplish the potential value of EOL product recycling
compact spaces, potentially leading to accidents. There- with the uncertainty of end-of-life products, which makes
fore, safety constraints are typically added in the planning the human–robot collaboration disassembly strategy need
of collaborative disassembly. With the continuous devel- to dynamically optimize in real time disassembly processes
opment of Industry 4.0, intelligent technologies such as or sequences. Rickli et al. [11] proposed the impact of
cyber-physical systems and cloud computing are gradually uncertain EOL product quality and the optimal disassem-
being introduced into the remanufacturing process. bly sequences to establish a correlation between the aging
Considering the requirements of addressing the uncertain- distribution and their quality, which introduced a
ty of EOL products and collaborative safety, digital mathematical framework to assess the optimal disassembly
twinning technology is gradually being introduced into the sequence by analyzing the partial disassembly sequence in
HRC disassembly process. The digital twinning technology age distribution of EOL products. The uncertainty of EOL
was first proposed by Professor Grieves in 2002 and refers products poses many challenges to the development of an
to the interaction of real-time digital models with physical optimal disassembly process planning system with the
entities (products, equipment, systems, or processes) to quality uncertainty of EOL products, which can be used to
monitor, analyze, and optimize their operational status. address the uncertainty by optimizing the disassembly
Digital twinning not only reflects the geometric features of process. Zhu et al. [12] developed a comprehensive
physical entities, but also encompasses their operational disassembly framework by combining an ontology-based
status, behavior, and performance. This technology information model, which simultaneously demonstrated
enables comprehensive simulation, predictive mainte- the possible optimization of self-adaptive disassembly
nance, and performance optimization of physical entities planning. Tian et al. [13] proposed a fuzzy perspective
to facilitate real-time decision supports. The applications approach to analyze the uncertainty of disassembly process
of digital twinning technology can identify potential that combines fuzzy simulation and artificial bee colony
hazards and risk scenarios to predict possible risk methods. Empirical evidence has shown that the algorithm
situations, which help to take preventive measures, thus can generate superior solutions through subsequent fuzzy
enhancing the safety of collaborative disassembly. Fur- simulations. Laili et al. [14] proposed an interference
thermore, the other decision-making technologies (i.e., probability matrix to represent the EOL product in the
reinforcement learning, neutral networks, etc.) [9] can disassembly process by establishing a multi-threshold
provide the specific solutions to combine HRC disassembly optimization of disassembly sequencing. The quality
with flexibility and efficiency. This paper aims to review defects of end-of-life (EOL) products in the disassembly
the relevant application techniques for multi-agent HRC process mainly include the wear, deformation, corrosion,
disassembly. The related literatures were retrieved from aging, etc. Meng et al. [15] proposed a multi-constraint
the Web of Science, Science Direct, and IEEE Xplore remanufacturing disassembly line balancing model to
databases, etc., including the journal papers and the address multiple failure and material hazard constraints,
conference papers according to the specific reviewing which integrates different engineering aspects (i.e., product
procedure as shown in Figure 1. failure modes and failure degrees), while demonstrating the
efficiency of remanufacturing disassembly line systems.
The uncertainty of product quality might cause the
2 Scientific challenges towards intelligent variability of disassembly time for disassembly tasks,
disassembly which causes the disassembly line balance problem (DLBP)
for EOL product recycling as shown in Table 1. It is
In the remanufacturing fields, intelligent robots are necessary to compare the various research points based on
working alongside human operators to perform remanu- various optimization and decision-making methods for
facturing tasks such as assembly, disassembly, and HRC disassembly process, including expected disassembly
CNC machining [10]. The continuous development of times, profit from disassembly, operation safety, and
 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024) 3

Web of Science Science Direct IEEE Xplore

Literature Database

Language: English; NO
Type: Research paper or
Conference paper.

YES

Literature categorization Exclude

Topic:
Scientific challenges NO
Automatic remanufacturing platform
Intelligent disassembly
technology

YES

Contrastive analysis

Discussion and Summary

Fig. 1. The procedure of reviewing the related journal and conference papers.

Energy consumption of disassembly equipment. To address effectively, Yin et al. [20] proposed a novel HRC flexible
the correlation between uncertain disassembly tasks, Liu remanufacturing system that enables practical battery
et al. [16] proposed a novel distributed robust analysis disassembly production lines.
model with joint multi-constraints to optimize disassembly
line balance. Zhang et al. [17] focused on energy 2.2 Disassembly dynamic structure
consumption to present the stochastic energy consumption
disassembly line balance problem (SEDLBP), which The diversity of disassembly products presents the
constructed a spatial interference matrix for SEDLBP significant uncertainty of the HRC disassembly process,
analysis perspective. Furthermore, Liu et al. [18] addressed thereby causing the challenging of complex disassembly
stochastic optimization in disassembly sequence planning tasks. As considering the variety of disassembly objects,
by proposing an improved peacock optimization algorithm dynamic disassembly is becoming as key scientific problems
to explore optimal disassembly sequences. By considering in the disassembly structure [31]. The disassembly
the requirements of EOL product recycling to enhance the representations such as Petri nets, AND/OR diagrams,
recovery efficiency and reduce the recycling costs, it and direct diagrams can be used to deduce the possible
becomes imperative to identify an appropriate recovery disassembly roadmap as shown in Table 2.
mode and HRC disassembly procedure. Xiao et al. [19] The comprehensive literatures were reviewed by
proposed a multi-agent reinforcement learning method for analyzing the specific research points based on the
retired electric vehicle (EV) batteries, which discusses the application methodologies. Singh et al. [32] enhanced the
potential for future hierarchical utilization and disassem- disassembly Petri Net model, and subsequently developed
bly of EV batteries toward uncertain disassembly and the Expert Disassembly Petri Net and the Expert
hierarchical application scenarios. To tackle this challenge Enhanced High-Level Coloured Disassembly Petri Net.
4 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024)

Table 1. Reviewing the related literatures for HRC disassembly.

HRC disassembly Ref. Research contents Main concerning research points

optimization methods

Expected Profit Operator Energy

disassembly from safety consumption
time disassembly of disassembly
equipment
*
p
A mathematical Rickli et al. [21] Analyze the expected profit of
optimization framework disassembled products.
based on standard deviation
and profit probability. * Determined optimal disassembly
sequence with GA algorithm.

*
p
A disassembly information Zhu et al. [22] Develop information models and
framework with ontology- computational models by using
based information model. product uncertainty.

*
p
A novel hybrid intelligent Tian et al. [23] Evaluate the uncertainty with
algorithm integrating fuzzy fuzzy simulation.
simulation and artificial bee
colony. * Determined the optimal disas-
sembly sequence with ABC algo-
rithm.

*
p p
A multi-threshold scheme Laili et al. [24] Established IPM to reflect uncer-
based on interference tainty in interference conditions.
probability matrix.
* Multi-threshold sequence plan-
ning using IPM.

*
p p
A novel multi-constraint Meng et al. [25] Construct MC-RDLB Mathema-
disassembly line balancing tical Model.
model. * Solve MC-RDLB by using
NSGA-II algorithm.

*
p
A new distributionally Liu et al. [26] Propose a distributionally robust
robust analysis with a joint joint chance-constrained formu-
constraint. lation.

* Propose Two-stage parameter-

adjusting heuristic.

*
p p
A mathematical model of Zhang et al. [27] Constructs mathematical model
SEDLBP based on spatial of SEDLBP based on the disas-
interference matrix sembly of spatial interference
disassembly. matrix.

* Solve SEDLBP by using stochas-

tic simulation.

*
p
An improved peafowl Liu et al. [28] Construct multi-objective math-
optimization algorithm. ematical model.

* Construct IPOA algorithm fra-

mework.
 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024) 5

Table 1. (continued).

HRC disassembly Ref. Research contents Main concerning research points

optimization methods

Expected Profit Operator Energy

disassembly from safety consumption
time disassembly of disassembly
equipment

*
p p
Hierarchical analysis Xiao et al. [29] Analysis technical standards for
methods based on various EV batteries Echelon Utiliza-
battery evaluation index. tion.

* Analyze the EV batteries disas-

sembly planning method.

*
p
A novel human–robot Yin et al. [30] Analyze the kinematics of colla-
collaborative flexible boration robot with D-H para-
remanufacturing system. meter method.

* Planning trajectory of collabora-

tion robot with RRT algorithm.

By considering the high-efficient disassembly points as an method that integrates conflict matrix, precedence and
example, these three methods were compared to demon- succession matrices to explore a feasible disassembly
strate their practicality in disassembly sequence planning. sequence. In addition, to address these dynamic disas-
Turowski et al. [35] proposed a Fuzzy Coloured Petri Net sembly problems, Zhu et al. [42] incorporated dynamic
model to characterize the influence of uncertain disassem- disassembly capabilities into the information model to
bly factors based on heuristic algorithms to balance the accommodate the disassembly operations, which subse-
disassembly lines. In their subsequent scientific research, quently discussed an optimization model. Ye et al. [43]
an adaptive fuzzy system was proposed by an enhanced integrated a fuzzy and dynamic modeling method with an
learning method for predicting environmental impacts [36]. iterative re-planning strategy to enhance the adaptability
Tang et al. [37] considered product uncertainty caused by of disassembly planning process. Furthermore, by explor-
human intervention in the disassembly process to represent ing artificial intelligence and machine learning as potential
the effects of membership functions, which addressed the solutions for these challenges, Meng et al. [44] synthesized
disassembly optimization by developing a fuzzy attributed key advancements and research opportunities in emerging
Petri net model. In their further research, Tang et al. [38] smart technologies for recycling and reuse of industrial
proposed an enhanced fuzzy learning algorithm with products, with a particular emphasis on resolving
variable memory length to ensure the robustness of uncertainty issues. Therefore, it is necessary to combine
adaptation processes. Similarly, Tian et al. [39] analyzed with the artificial intelligence and with HRC disassembly,
the uncertainty of evaluation algorithms for disassembly which can effectively address such uncertainties and
indicators with the fuzzy sets to represent the structure significantly improve the adaptability of disassembly
information, thereby transforming Petri net models into processes.
fuzzy analysis method. Therefore, the different character-
istics of disassembly optimization methods can be
compared as shown in Table 3. 3 Automatic disassembly in remanufacturing
However, Mendez et al. [40] proposed the direct and
indirect graph to represent the product structure that With the continuously increasing volumes of End-of-Life
determines the optimal disassembly sequence. Edmunds (EOL) products, traditional labor-intensive manual disas-
et al. [33] proposed a constraint-based method to sembly is no longer able to meet the efficiency requirements
investigate the optimal disassembly sequences by trans- for disassembly towards semi-automated disassembly.
forming the AND/OR structure into a precedence Semi-automated disassembly, compared to manual disas-
structure. Tian et al. [41] highlighted that the existing sembly, offers multiple advantages including disassembly
heuristic disassembly planning to adequately address the efficiency, cost reduction, disassembly safety, and environ-
structure relationship with a triple-phase adjustment mental protections. With the sustainable development of
6 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024)

Table 2. The comparison between disassembly structure representations and their characteristics according to the
related papers.

Disassembly structure References. Advantages Disadvantages

representation
Petri Net Singh et al. [32] Effective concurrent behavior Complexity in modeling and
and resource competition. analysis for complex
simulation and quantitative scenarios.
analysis. Limited capability.
AND/OR Graph Edmunds et. al. [33] Clear representation of task A large structure in complex
dependency. scenarios.
Relatively intuitive and Lack of capability in resource
simple implementation. constraints.
Direct Graph Mendez et al. [34] Clear representation of Limited capability
disassembly task sequential Ambiguity and complexity
relationship.
Easily transformed into
execution paths.

Table 3. The different characteristics of disassembly optimization methods.

Optimization methods Colour identification Influence Probability Average profit Operation time
DPN
p p
EDPN
p p p
EEHLCDPN
p p p p
FCPN
p p p
FAPN
Note: DPN/ Disassembly Petri Net; EDPN/ Expert Disassembly Petri Net; EEHLCDPN/ Expert Enhanced High-Level Coloured
Disassembly Petri Net; FCPN/ Fuzzy Coloured Petri Net; FAPN: Fuzzy attributed Petri net.

modern remanufacturing fields, collaborative robots, as a bly experiment with a camera as the object. Hohm et al.
core tool of semi-automated disassembly, play a crucial role [48] explored the feasibility of robot-assisted disassembly
on in automation, precision operations, data collection, and on standard PC hardware, which focused solely on opened
analysis. Therefore, it is necessary to provide a review PCs for the disassembly process. Wegener et al. [49] devised
and analysis of robot-assisted disassembly technology and a battery disassembly workstation to disassemble the EOL
human–robot collaboration disassembly technology as batteries from electric vehicles, focusing only simple and
shown in Figure 2. repetitive disassembly tasks with robot operations, while
the complex disassembly tasks still relied on skilled human
3.1 Robot-assisted disassembly operators. The manual disassembly in the semi-automatic/
automatic disassembly production line should be imple-
To achieve the autonomous disassembly based on robot- mented gradually. The automatic disassembly of industrial
assisted remanufacturing, it is necessary to reduce the EOL products is typically significant to demonstrate the
current manual disassembly owing to its time-consuming economic performance to manual disassembly [50]. In
and lower efficiency of disassembly. Moreover, the hazardous disassembly and recycling process, it can reduce
disassembly of certain EOL products may cause the the dependence on labor-intensive human works to
generation of hazardous materials, which is an inevitable improve the efficiency of EOL product recycling [51].
trend for robots to assist the disassembly production lines. Because the uncertainties of current fully automated
To address these challenges, Billard et al. [45] discussed the disassembly line development are more difficult, the
possibility of robot and human operators together to current subsequent disassembly technology should be
manipulate disassembly objects. Caiza et al. [46] consid- more attention on the development of intelligent collabo-
ered the remote control to accomplish the disassembly rative disassembly. Foo et al. [6] demonstrated that there
operations of robot as the operators to minimize the exists a significant correlation between the automation
potential operation risks. The robot-assisted disassembly level and the disassembly capability to solve uncertain
has commenced to develop the gradual advancement for disassembly while exploring cognitive robots and semi-
automatic production line as shown in Figure 3. Weigl et al. automatic disassembly. By reviewing relevant literature
[47] emphasized the significance of robot-assisted disas- published in the past decade, it is necessary to compare the
sembly in EOL product recycling, conducting a disassem- number of journal or conference papers related to human–
 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024) 7

13%
27% 27% 29%

9% 29%

37% 29%

Simple Disassembly Task Optimziation Disassembly Safety

(Wegener et al. 2015) (Lee et al. 2020; Lee et al. 2022; Liao et al. 2022)
Safety Analysis Human Behavior Prediction
(Bogue et al. 2019; Caiza et al. 2020; Chen et al. 2023) (Lee et al. 2023; Tian et al. 2023)
Uncertainty Optimization disassembly Time Optimization
(Weigl et al. 1994; Vongbunyong et al. 2013; Foo et al. 2022) (Xu et al. 2020; Liao et al. 2022; Guo et al. 2023
Disassembly Quality Analysis Uncertainty Analysis
(Hohm et al. 2000; Tolio et al. 2017; Li et al. 2018; Liu et al. 2020) (Liu et al. 2019; Liu et al. 2019; Liao et al. 2022

(a) (b)
Fig. 2. Statistical proposition of (a) robot-assisted disassembly (b) human–robot collaboration disassembly in the remanufacturing
process.

robot collaborative disassembly. In fact, in 2019, the which enables real-time identification of item locations,
number of papers specifically focused on human–robot direction for disassembly, and operator positioning for
collaborative disassembly exceeded those related to robot- optimal disassembly planning [56]. During the entire
assisted disassembly. This finding demonstrates that the disassembly process, the disassembly operation action
greater potential of human–robot collaborative disassem- may also result in frequent interruptions to the robot,
bly can be used to improve the efficiency, safety, and thereby reducing overall disassembly efficiency. Accurate
economic benefits of the disassembly processes. disassembly behavior can minimize the robot idle time to
consequently enhance the efficiency of disassembly. Lee
3.2 Human-robot collaboration disassembly et al. [57] considered that disassembly operation actions
would interfere with robot actions by a sequence planner to
The disassembly process towards uncertain disassembly predict the potential human actions that can be used to
objects has many intricate and diverse disassembly optimize task allocation by considering the human actions
operations by considering the applications of HRC for and real-time product status. Ding et al. [58] developed a
safer, more efficient, and time-saving disassembly oper- knowledge-based platform for storing disassembly data,
ations [52]. With the gradual capabilities of robot which provides the disassembly methods for robots to
perception, HRC disassembly has become a feasible reduce the human workloads that helps reduce robot idle
method to support the intelligent disassembly. Liu et al. time and improve overall efficiency. Tian et al. [59]
[53] proposed a hybrid HRC disassembly framework that proposed a rational prediction model to analyze human
encompasses the perception, cognition, and decision- behavior and motion based on an HRC dataset of
making to address the uncertainties and dynamics disassembly tasks with the optimal disassembly processes.
relationships in the disassembly process. The feasibility To reduce the disassembly time of EOL products, it is
of disassembly system has been validated through subse- necessary to improve the overall disassembly efficiency,
quent case studies, which also integrate the advanced which involves the balance of HRC disassembly lines for
technologies such as cyber-physical production system reasonable task allocation with optimal disassembly
(CPPS), artificial intelligence (AI), etc. The researchers sequence. Xu et al. [60] proposed an enhanced discrete
further presented a machining learning-based method for bees algorithm (MDBA-Pareto) based on Pareto optimi-
predicting human motion using visual signals without any zation to determine the optimal solution for all feasible
specialized wearable devices while still enabling effective disassembly sequences. Guo et al. [61] developed an AND/
human–robot interaction based on Digital Twin (DT) [54]. OR structure graph to demonstrate the disassembly
Although HRC disassembly effectively handles the uncer- process combining with a HRC disassembly task times,
tainty in the disassembly process, the human operator can which addressed it using a Pareto improved multi-objective
impact the robot actions when compared to fully automatic algorithm. To further enhance the performance of
disassembly. Lee et al. [55] proposed a disassembly disassembly optimization algorithm, it is necessary to
sequence planner for human operator safety with multiple optimize the disassembly strategy into the global disas-
constraints to co-robots, which demonstrated that the sembly process. However, the future HRC disassembly is
robot can successfully complete its disassembly tasks to expected to enhance the intelligence of collaborative
safety constraints through a simulated disassembly robots, thereby improving the efficiency and safety of
experiment. Subsequently, a real-time disassembly the entire disassembly process. Similarly, there are still
sequence planner also prioritizes the disassembly safety, significant potentials for research on HRC disassembly
8 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024)

Fig. 3. Comparison of the number of related published literatures.

strategies, with artificial intelligence technology and digital 4.1 CPS for disassembly
twin technology potentially to accomplish the new break-
throughs [62]. Cyber-physical system (CPS) is a complex system that
connects physical devices to the Internet, enabling them to
perform the complex calculations, communication, precise
4 Intelligent disassembly technology towards control, remote coordination, and production management
Industry 4.0 by computing and control technologies. This enables real-
time perception, information services, and dynamic control
The development of Industry 4.0 or even Industry 5.0 has of large-scale systems by integrating the virtual network
brought profound changes and opportunities with the world with the physical world as shown in Figure 4. A new
intelligence and human-centric considerations to the rema- generation of intelligent systems can be built to integrate
nufacturing industry. Firstly, the intelligent manufacturing the computing, communication, and control functions of
technology has caused the transition from traditional manual CPS-based remanufacturing technology [63]. The informa-
operations to digital and automated production, thereby tion physical systems in HRC disassembly have been
enhancing the efficiency of product disassembly in the explored the intelligent disassembly process. Kopacek et al.
remanufacturing process. Specifically, the applications of [64] emphasized the application of intelligent systems to
sensors, robots, and decision-making systems in the remanu- facilitate communication and collaboration of different
facturing process enables more precise and efficient product entities during the disassembly process. Additionally, CPS
disassembly. In addition, the integrated application systems systems can monitor the entire HRC disassembly oper-
such as the Internet of Things, big data, and artificial ations and generate virtual replicas to optimize the
intelligence has brought new opportunities for digital remanufacturing task that can be used to provide the
management and intelligent production to the remanufactur- decision-making and improve the efficiency of EOL product
ing industry. By considering the real-time monitoring, data disassembly.
analysis, and predictive maintenance, remanufacturing The CPS-based disassembly system can be divided into
enterprises can better manage production processes by three various layers, including the perception layer, the
optimizing the production planning and reducing production network layer, and the control layer [65,66], which are
costs, which improve the efficiency of resource utilization or interdependent with each other. As shown in Figure 4, the
recycling. Furthermore, intelligent remanufacturing systems perception layer primarily consists of sensors, controllers,
based on big data and artificial intelligence can autonomously collectors, and other devices, which can be used to collect
identify and evaluate the feasibility of EOL product information and data from the environment and transmit
disassembly, while providing more intelligent support for them to the server with the real-time perception of the
decision-making in the remanufacturing process. The intelli- disassembly process. Nikolakis et al. [67] developed a cyber-
gent disassembly technologies have also facilitated the physical context-aware system that monitors production
sustainable remanufacturing industry as a key element of lines that facilitates HRC manufacturing process by
the circular economy by extending product lifespans and gathering the real-time data. Filipescu et al. [68] proposed
recycling the resources and materials. a hybrid assembly/disassembly production line using an
 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024) 9

Cyber-Physical
System

The Perception The Network The Control

Layer Layer Layer

Sensors Wired/Wireless Deep Learning

Controllers Network Machine Learning
Collectors Devices 

Real-time
Product Personnel Exchange of Disassembly Disassembly Safety
Robot Status Data
Information location Transmission
Information sequence tasks warnings

(Nikolakis et al. 2018; (Lee et al. 2008; (Nikolakis et al. 2018; Kopacek et al. 2019;
Kopacek et al. 2019; Hahn et al. 2015; Liu et al. 2019; Jacomini Prioil et al. 2020;
Filipescu et al. 2023) Alguliyev et al. 2018) Filipescu et al. 2023; )

Fig. 4. The different application layers about the disassembly and CPS.

autonomous cyber-physical system (ACPS) for recycling of disassembly process of EOL products. Radio Frequency
the EOL products. In the multi-layer architecture, the devices Identification Tags (RFID) are widely used to transfer the
such as monitors are used to implement the functions of the device data within the IoT that is capable of being
sensing layer. The network layer is used to transfer the real- identified through contactless frequency signals. By
time monitoring data, which establishes the connection reviewing the comprehensive literatures, Ferrer et al.
between the information world and the physical world by [75] extensively discussed the application of RFID
achieving device interconnection and interoperability. The technology in remanufacturing processes to demonstrate
application control layer primarily focuses on real-time its potential value that can be used to simplify the EOL
analysis and processing of data by combining the correspond- product disassembly and enhance the overall efficiency of
ing intelligent algorithms to accomplish the disassembly tasks disassembly process. The application of the IoT combined
such as sequence planning optimizations and real-time with HRC disassembly process can be demonstrated as
decision-makings. Liu et al. [53] proposed a novel concept shown in Figure 5.
of Cyber-Physical Production Systems (CPPS) into a HRC Before disassembling the EOL products, the pre-
disassembly framework to acquire the intelligent decision- processing is required to collect characteristic data such
makings based on reinforcement learning that can be used to as the size and structure of EOL products for subsequent
achieve efficient collaborative disassembly tasks between disassembly planning and decision-making of HRC disas-
humans and robots. Prioli et al. [69] proposed a notion of a sembly processes. The related data can be used to classify
cyber-physical information system to develop a training the similar EOL products, which can enable the application
module to enhance the flexibility of disassembly process with of a feasible disassembly planning to the EOL products [76].
co-robots. In addition, it will be meaningful to further explore The IoT can play a significant role in the pre-processing
the research on Human-Cyber-Physical Systems (HCPS), process by recording the data label by the EOL product to
which can be seen as an extension of the information physical obtain basic disassembly information. Wessel et al. [77]
system that incorporates human as a key unit into the overall proposed a tracking and tracing system based on
manufacturing or remanufacturing system. This integration morphological analysis that provides an automatic identi-
primarily aims to leverage the human cognition and their fication system based on IoT technology to acquire the
specific actions to optimize the functionality across different product information. Kumar et al. [78] proposed the
application systems [70,71]. Similarly, this HCPS disassembly product information to combine the IoT with artificial
framework primarily aims to enhance the flexibility of intelligence methods, which determine the near-optimal
application systems [72] to ensure the safe of HRC disassembly disassembly sequences by validating the collected product
[73] to balance the disassembly cost and time [74]. information to identify any missing data sets and the
operation errors during the disassembly process [79]. The
4.2 IOT for disassembly EOL products also require a preliminary assessment of their
disassembly condition, while encompassing the health status
Disassembly is a crucial step in the remanufacturing of EOL products, degree of damage, and other pertinent
process combining with the emerging technologies like the information [80]. By considering the HRC disassembly of
Internet of Things (IoT). However, it is necessary to retired EV batteries as an example, it is generally imperative
overcome certain challenges encountered during the HRC to analyze the characteristics such as remaining battery
10 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024)

IoT platform

Product Develop Collaboration

... Pretreatment Disassembly Plan Disassembly ...

Identify: Combine: Monitoring:

RFID Artificial Intelligence Disassembly Progress
QR code Cloud Computing Equipment Status
... ... ...

Obtain Information: Output: Scheduling:

Size and Structure Disassembly sequence Collaborative robots
Health Status Task Allocation Resource Allocation
... ... ...

(Ondemir et al. 2014; Yi et al. 2015;

(Wessel et al. 2020) (Kumar Gulivindala et al. 2021)
Zhang et al. 2018)

Fig. 5. The framework of the IoT applications about the disassembly process.

capacity and the hazardous substances before disassembly process and customer service [3]. Furthermore, it is
operations [81]. However, it is crucial to ensure the necessary to combine the cloud-based technologies such
subsequent disassembly safety during the disassembly as cloud computing and IoT technology that can integrate
process. The IoT system can also be utilized for real-time the useful information or knowledge for future collabora-
monitoring and scheduling of the collaborative disassembly tive disassembly technology [78].
process. Ondemir et al. [82] proposed a monitoring and data Due to the uncertainty of EOL products, the pre-
sharing framework for quality management and process processing of disassembly plays a crucial role in stream-
optimization in EOL product recycling to improve the lining disassembly operations, while cloud-based remanu-
efficiency of the recycling processes. Yi et al. [83] demon- facturing systems can also contribute significantly to the
strated that IoT technology can be used to explore the process. Xia et al. [85] integrated the Q-learning algorithm
recycling of EOL products based on RFID tags to match and with a cloud-based remanufacturing system to propose the
track the retired vehicles or its originals. The disassembly diverse disassembly multi-objective optimization. Caterino
status and disassembly processes of EOL products in the et al. [3] proposed a cloud remanufacturing system to
disassembly production are also transmitted to the real-time collaborate with initial information screening for the
IoT system to support the remote monitoring. Zhang et al. disassembly process. However, manufacturer-related in-
[84] establish an IoT manufacturing environment to achieve formation can be analyzed by screening the related data
real-time production scheduling for the entire remanufac- from the disassembly planning stage that can be used to
turing system by capturing disassembly information, which reduce planning costs and simplify product disassembly
provides the experimental verification to improve the operations. The cloud-based remanufacturing technology
effective scheduling. can enhance the efficiency of resource scheduling during the
disassembly process. Jiang et al. [86] proposed a cloud-
4.3 Cloud-base remanufacturing system for based disassembly system that can accomplish the
disassembly disassembly capabilities of various factories to support
different disassembly tasks into disassembly resources.
With the development of intelligent manufacturing These resources can be allocated and scheduled based on
technologies, cloud technology and other supporting specific disassembly requirements to enable the personal-
related technologies are gradually being explored in the ized disassembly services. However, it is necessary to
remanufacturing fields, offering a novel approach for data combine the HRC disassembly processes, which can be
analysis and information exchange platform in the executed by the disassembly operators with multiple
intelligent production process as shown in Figure 6. disassembling robots. Each robot can be assigned according
However, it is important to enhance the integration of to different disassembly operations or tasks, which can
cloud technology, which provides the possibility of HRC allow the real-time scheduling through cloud-based
disassembly operations in the remanufacturing process. remanufacturing system to meet diverse requirements of
However, the implementation steps of remanufacturing EOL product disassembly. Mancusi et al. [87] proposed a
activities can utilize the cloud-based remanufacturing cloud computing technology to analyze the data from
platform to establish a connection between various disassembly process by collecting storing, mining, and
information systems regarding the remanufacturing transmitting to support the disassembly decision-makings.
 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024) 11

(Xia et al. 2014;

Store disassembly (Mancusi et al. 2022)
Wang et al. 2014;
process data
Caterino et al.
2020)

Reading Writing

Option Status

Cloud Computing

Requirement Task

Data Strategy
Meet the disassembly Resource sharing and
needs of customers collaborative cooperation

(Kumar Gulivindala (Wang et al. 2014;

et al. 2021; Caterino Big data analytics
and intelligent Jiang et al. 2016;
et al. 2022; Mancusi decision-making Zhang et al. 2015)
et al. 2022)

Fig. 6. The fundamental framework of cloud computing technology.

4.4 AI-base remanufacturing system for disassembly working station, safety status within the complex environ-
ments by avoiding obstacles or hazards during safe
AI-based remanufacturing systems can fulfill various roles disassembly operations.
to improve the efficiency, safety, and precision of After reviewing loads of literatures, it has been
disassembly operations. The subsequent applications of summarized that genetic algorithms, artificial ant colony
artificial intelligence technology can be divided into optimization, and artificial bee colony algorithms have
disassembly sequence planning, disassembly line balance, more attentions on the research of optimization and
and related image recognition and object detection in decision-makings. Furthermore, by analyzing the algo-
disassembly process. The disassembly sequence optimiza- rithm characteristics according to different disassembly
tion algorithm (i.e., reinforcement learning, heuristic objects, it is necessary to optimize the specific disassembly
algorithms, etc.) can effectively optimize the disassembly sequencing or tasks in the real disassembly operations.
sequence for EOL products while considering multiple Kheder et al. [89] analyzed the probability of disassembly
disassembly conditions such as product structure, disas- sequence planning, which provides the optimization
sembly time and cost as shown in Table 4. However, it is algorithm to converge on local optimal solutions. With
possible to optimize the disassembly operation time, the applications of robots and human operators in the HRC
manpower allocation, and resource utilization within the disassembly production line, the effective disassembly task
disassembly line. Through computer vision technology, allocation has become crucial for achieving the intelligent
artificial intelligence combined with robot agents can be and more efficient disassembly. Ranz et al. [120] proposed a
used to identify various positioning of disassembly objects reasonable disassembly task allocation to ensure the
by tracking their position and operation status. This optimal solution of HRC disassembly by leveraging their
vision-based information supports robot-assisted disas- respective advantages, thereby improving overall produc-
sembly operations by combining computer vision with tion efficiency. Moreover, by optimization the disassembly
deep learning technology. Robots can identify obstacles, strategies, the potential capabilities of both humans and
12 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024)

Table 4. The comparisons between various disassembly optimization methods and disassembly characteristics by
reviewing the related literatures.

Disassembly optimization methods References. Disassembly sequence Disassembly Scenario

optimization line balance recognition
and object tracking
p p
Go et al. [88]
p
Genetic Algorithms Kheder et al. [89]
p
Tseng et al. [90]
p p
Lu et al. [91]
p
Ant Colony McGovern et al. [92]
p p
Hu et al. [93]
p p
Liu et al. [94]
p
Hartono et al. [95]
Artificial bee colony p
Kalayci et al. [96]
p p
Zhang et al. [97]
p p
Pornsing et al. [98]
p
Tseng et al. [99]
Particle swarm optimization p p
Tao et al. [100]
p
Kalayci et al. [101]
p p
Alshibli et al. [102]
p
Tao et al. [103]
Tabu search p p
Kalayci et al. [104]
p
Liang et al. [105]
p
Xia et al. [106]
p
Simulated annealing Fang et al. [107]
p
Wang et al. [108]
p
Chu et al. [109]
Reinforcement learning p
Allagui et al. [110]
p
Yildiz et al. [111]
p
Convolutional neural networks Li et al. [112]
p
Adesso et al. [113]
p
Brogan et al. [114]
p
YOLO algorithms Mangold et al. [115]
p
Zhang et al. [116]
p
Recurrent neural network Zhang et al. [117]
p
Deng et al. [118]
Long short-term memory p
Chen et al. [119]

robots can be maximized to provide the greater flexibility process by controlling the optimization algorithm. In
and adaptability of the disassembly process. The reason- addition, it is necessary to enrich the dataset for
able disassembly task allocation reduces the risks of HRC disassembly processes to enhance algorithm accuracy
disassembly operations while ensuring safety and reliabili- and generalization capabilities, which can be used to
ty of the disassembly system. Chen et al. [121] combined facilitate seamless integration between artificial intelli-
genetic algorithms (GA) for real-time disassembly task gence technology and remanufacturing systems to improve
assignment with the effective requirements reliably. Wu the efficiency of HRC disassembly systems.
et al. [122] proposed a trust-based dynamic disassembly
task assignment method for HRC system to enhance the 4.5 Digital twin-driven multi-agent disassembly
information interoperability of the disassembly system.
Owing to the diversity and complexity of disassembly The implementation of digital twin technology in the semi-
objects and structures, it is necessary to ensure the effective automated disassembly process facilitates collaborative
safety of the actual disassembly during the disassembly disassembly of human operators and robots by generating a
 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024) 13

(Verna et al. 2023χ

Li et al. 2023)

Artificial intelligence:
• Genetic Algorithm
• Deep Q-learning
(Liu et al. 2023; • ... (Kerin et al. 2023;
Qu et al. 2023) Mouflih et al. 2023)

Disassembly process: Output:

• Product Information Enrich Adjust • Disassembly sequence
• Robot States Virtual model • Task Allocation
• Operator Position Simulation • Operator Position
• Environmental Information • Equipment States
• ... • ...

• Internet of Things
(Mouflih et al. 2023χ
• Knowledge database
Baratta et al. 2024)
• ...

Optimize

Fig. 7. The application of digital twin-assisted disassembly platform.

virtual replica of the physical system. Through this determine the optimal disassembly path. Verna et al. [125]
technology, the data and information can be shared for combined digital twin technology with the prediction
multi-agents within the virtual environment, which can models to ensure optimal disassembly planning and quality
collaborate effectively to accomplish the intricate disas- control. Liu et al. [9] analyzed digital twin technology for
sembly tasks. The disassembly process can be simulated identifying missing conditions in product disassembly for
and optimized based on digital twin platform that can constructing virtual disassembly spaces and enriching
enhance the efficient execution of disassembly operations. digital twins-based perceptual data transmission. Similar-
The overall application mode of digital twin technology in ly, the reinforcement learning algorithm can be used to
human–robot collaboration disassembly process can be dynamically generate an optimal disassembly sequence
demonstrated as shown in Figure 7. As reviewed many while leveraging input data based on digital twin platform
literatures, Mouflih et al. [123] proposed the potential role to validate and optimize the disassembly strategy genera-
of digital twins in the disassembly process, which provide tion. Digital twin technology also provided the capabilities
the decision-making information during product disassem- of the simulation and analysis in different spatial and
bly to address the efficiency of information exchange. temporal scales to simulate and analyze the behavior of
Baratta et al. [124] focused their research on the systems with the temporal data and information. However,
advantages of digital twins in HRC system, which proposed the multi-scale digital twin technology can simulate and
the robust capabilities of digital twins in simulation testing optimize the HRC disassembly processes from individual
and optimization of collaboration strategies. However, workstations to entire disassembly production lines to
digital twin technology plays a crucial role in multi-agent improve the efficiency and safety of the disassembly
collaboration and disassembly processes to offer new process. Kerin et al. [126] proposed the digital twinning of
possibilities for optimizing the disassembly process. the EOL products by sensors and IoT technologies to
When multiple agents collaboration accomplishes the capture product information in real-time or near real-time
disassembly tasks for EOL products, the decision-making simulation process, thereby accurately replicating the
of disassembly process might be affected by uncertainties virtual information of the modeled product. Mouflih et al.
such as the damage of the EOL products. Digital twin [127] considered the digital twin technology of EOL
technology enables the real-time monitoring and analysis of products to support sustainable disassembly decision-
the EOL product status, while predicting potential issues makings. Because the virtual model can integrate a large
that may cause during disassembly processes. Further- amount of data from EOL product knowledge base, Qu
more, digital twin technology facilitates the optimization of et al. [128] developed a digital twin platform of the HRC
disassembly planning and control in the disassembly disassembly environment to reflect the real-world data
process by simulating various disassembly process to into the virtual space for simulation, monitoring, and
14 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024)

Table 5. Representative literatures about the actual industrial disassembly objects.

References. Disassembly object Optimization method Safety constraints

Xiao et al. [8] Electric Vehicle Battery Multi-Agent Reinforcement Learning No
Liu et al. [9] Camera\Gear pump Digital twin\Deep Q learning network No
Liu et al. [54] Desktop computer Convolutional neural network\Long short-term Yes
memory network
Lee et al. [56] Toy box Real-time receding horizon sequence planner Yes
Guo et al. [61] Hammer drill\ A Pareto improved multi-objective shuffled No
Copying machine\ frog leaping algorithm
Radio set

optimization with intelligent algorithms and machining improvement and safety considerations in human–robot
learning methods to improve the efficiency and flexibility of collaboration disassembly to better practical industrial
the disassembly process. Li et al. [129] proposed digital disassembly.
twinning of the entire remanufacturing workshop to create In the case of human–robot collaboration disassembly
a virtual data to the real workshop, which enables real-time of EV batteries, the actions of both human operators and
virtual simulation and optimization. The digital twinning robots can be considered as the smallest unit of action
technology in multi-agent HRC disassembly is of signifi- related to the execution of a disassembly task. The
cant importance for achieving intelligent and efficient disassembly workstation can be conceptualized as a system
disassembly process. However, multi-Agent collaborative with the human operator and the robot [131]. Based on
modeling and simulation can be accomplished by digital observed EV battery structure, we can deal with specific
twinning technology to model and simulate multi-agent intelligent optimization techniques, the real-time optimal
systems to achieve simulation analysis and optimization of or near-optimal disassembly strategies and task allocation.
complex disassembly processes. In addition, intelligent With further development of HRC disassembly technology,
decision support systems can develop the intelligent its actual industrial application will expand to more
decision support systems based on digital twinning possible industrial application fields. For example, the
technology for the decision-makings of multi-agent HRC disassembly and recycling of automotive engine compo-
disassembly processes. Furthermore, intelligent monitor- nents can be explored by disassembling its valuable
ing and prediction can be constructed by digital twinning components in the engine, such as crankshafts, connecting
technology to achieve real-time monitoring and prediction rods, and gears, are typically high-quality metal products.
of multi-agent HRC disassembly processes. Finally, Similarly, the disassembled components can allow for the
intelligent optimization and resource allocation can be recovery and reuse of valuable materials, reducing resource
accomplished by digital twinning technology to intelli- waste and environmental burden, which also contributes to
gently optimize the workflow and resource allocation of the promotion of a circular manufacturing economy.
multi-agent HRC disassembly to achieve optimal disas-
sembly process. 6 Discussion
5 Actual industrial disassembly applications Through analyzing the existing scientific technologies, the
artificial intelligence and digital twin technology in the
By considering the varying levels of difficulty in the actual human–robot collaboration (HRC) disassembly process is
industrial disassembly processes, it is necessary to collect gradually paid more attention to consider the uncertainty
the engineering practices from representative literatures as of EOL products. Subsequently, robot-centered remanu-
shown in Table 5. In terms of engineering applications, facturing systems can accomplish the higher efficient
there is relatively more research focused on the disassembly disassembly in the disassembly process. However, the
of retired lithium-ion EV batteries. Due to the rapid growth disassembly method has brought about new issues such as
of electric vehicles in recent years, it is imperative to find disassembly sequence planning, disassembly task alloca-
out the solutions for the environmental pollution caused by tion, and collaborative safety. Furthermore, the applica-
recycling the lithium-ion batteries and its disassembly tions of advanced remanufacturing technologies (i.e.,
production [130]. In addition, we specifically investigated cyber-physical systems, Internet of Things, cloud comput-
whether these optimization methods included safety ing, artificial intelligence, and digital twin technology)
constraints, and over 50% of the studies considered this have more feasibility when facing with HRC disassembly.
aspect. However, some researches only focused on However, multi-agent human and robot collaboration
improving efficiency and flexibility while neglecting the in the context of disassembly has huge potentials to deal
safety aspect. Safety issues are crucial in industrial with the complex disassembly. The collaborative work with
applications, and determine the future massive applica- multiple robots can involve mutual cooperation to jointly
tions. Therefore, it is necessary to prioritize both efficiency achieve a disassembly goal or mutual assistance in different
 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024) 15

manufacturing processes. Through multi-robot collabora- after human–robot collisions can be efficiently predicted
tive disassembly, the flexibility and efficiency of the to ensure the safety of the entire disassembly process.
disassembly production line can be further enhanced to Through the simulation of virtual human behavior using
reduce the workload of manual operations. digital twinning technology, the effective prediction of
– Task allocation and planning: it is necessary to allocate human behavior can be achieved to avoid human–robot
disassembly tasks to different robots, maximizing their collisions. If a collision cannot be avoided, the simulation
capabilities and resources. Task allocation and planning can be used to find out the optimal robot working
algorithms should consider workload balancing, task feedbacks (such as returning to the original position,
interdependencies, and disassembly strategies for HRC. moving away from the operator, or shutting down
Similarly, IoT and cloud computing technologies can be directly) to achieve an effective balance between safety
utilized to dynamically allocate and optimize disassem- and efficiency.
bly tasks through digital twin platform. Real-time data – However, it is necessary to explore the integration of
collected by physical devices and sensors can be analyzed digital twinning technology with artificial intelligence to
by using cloud computing to provide optimal disassembly achieve the adaptive and intelligent disassembly deci-
task allocation strategies. sion-makings. The intelligent decision support systems
– Collaborative decision-making and control: it is neces- within digital twinning technology can is gradually
sary to achieve collaborative decision-making and developing intelligent decision-making capabilities for
control between human workers and multiple robots, different disassembly scenarios through the continuous
ensuring efficiency and safety in the collaborative optimizations of disassembly tasks. Furthermore, the
disassembly process. By considering the AI techniques, combination of digital twinning technology with artificial
such as deep learning and reinforcement learning intelligence can also facilitate the prediction and early
algorithms, collaborative decision-making and control warnings of potential disassembly issues during the
between multiple robots and human workers should be disassembly process. By integrating the risk prediction
further explored. By analyzing large-scale data and real- and fault diagnosis modules into digital twinning
time feedback data, intelligent HRC disassembly can be platform, the potential safety hazards can be dynamical-
enhanced, resulting in efficient collaborative decision- ly monitored in the real-time intelligent disassembly
making and control. system that can provide the timely safety precautions to
– Performance evaluation and optimization: it is necessary handle the recommendations for disassembly operators.
to evaluate the performance of multi-agent HRC – In addition, the real-time monitoring and feedback
collaborative disassembly through the use of defined systems of digital twinning technology in HRC disas-
performance metrics, simulation experiments, and data sembly processes can be used to enhance the efficiency
analysis techniques. Real-time optimization strategies and accuracy of the disassembly operations by the real-
can be implemented to enhance the efficiency and quality time calibration and comparison of the original model.
of disassembly objects. By combining with digital twin The actual data can enable the dynamic monitoring and
and big data analytics, it will be a trend to integrate real-time control of collaborative disassembly processes,
virtual model simulations, analysis of large-scale data, which can facilitate the timely detection of anomalies
extraction of valuable performance indicators, and during the disassembly process. Moreover, by feedback-
subsequent multi-objective optimization, etc. ing the actual data to the digital twinning platform, it
can analyze and evaluate the effectiveness and risks of
Meanwhile, the implementation of digital twinning the current disassembly method, as well as potential
technology in existing industrial disassembly lines may not optimization opportunities. Additionally, digital twin-
be widespread. With the increasing number of End-of-Life ning technology can be combined with expert systems
(EOL) products, the efficient requirements of HRC and knowledge graphs to achieve intelligent analysis and
disassembly will become increasingly stringent. As digital diagnostics of real-time monitoring data. By considering
twinning technology continues to integrate deeply into multi-domain knowledge and experiential rules embed-
intelligent disassembly, its application in future HRC ded in the complex disassembly model, the real-time data
disassembly will become more widespread, leading to many can be intelligently analyzed and assessed to provide the
potential research directions, including but not limited to: relevant optimization recommendation.
– The optimization of the digital twinning technology can – The application of digital twinning technology with other
achieve more accurate simulation and prediction of HRC intelligent technologies (i.e.: Internet of Things (IoT),
disassembly processes. The construction of digital cloud computing, etc.) can be used to promote the
twinning platform needs to consider complex character- innovation of the intelligent disassembly. Through the
istics such as multi-scale and multi-physics simulations real-time data collected by IoT devices, the digital
to reflect the models the accurate characteristics and twinning platform can capture the real-time physical and
behavior of the disassembled objects or disassembly virtual disassembly data by the collection methods,
production lines. By optimizing the established digital facilitating dynamic monitoring and intelligent decision-
twinning platform, the precise simulation and prediction making in the disassembly process. Combined with cloud
of the disassembly process can be carried out to enhance computing technology, it can enable the cross-regional
the accuracy of the output results. By focusing on collaboration, the large-scale data processing, and the
the existing HRC disassembly-related field, human real-time data feedback. The application of digital
operator behavior and emergency response measures twinning technology will further push the development
16 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024)

of intelligent disassembly to achieve the intelligent, Author contribution statement

efficient, and sustainable disassembly processes that can
provide the better technological support for the indus- Jinhua Xiao: Conceptualization, Methodology, Investigation,
trial disassembly productions. Formal analysis, Software, Writing  Original Draft, Writing 
By completely reviewing the optimizing HRC disas- Review & Editing. Kaile Huang: Data curation, Visualizations,
sembly mentioned in the paper, the general method Writing  Original Draft.
provided in the disassembly-related researches may not
be applicable to the specific disassembly lines, but these References
current challenges will be given attention on the future
researches to push the development of the flexible and 1. M.-L. Lee, X. Liang, B. Hu, G. Onel, S. Behdad, M. Zheng, A
intelligent disassembly process in the remanufacturing review of prospects and opportunities in disassembly with
fields. human-robot collaboration, J. Manuf. Sci. Eng. (2023) 1–26
2. Green Manufacturing: Fundamentals and Applications,
Springer US, Boston, MA, 2013
7 Conclusion 3. M. Caterino, M. Fera, R. Macchiaroli, D.T. Pham, Cloud
remanufacturing: remanufacturing enhanced through cloud
In this paper, the comprehensive reviews on human–robot technologies, J. Manuf. Syst. 64 (2022) 133–148
collaborative (HRC) disassembly will be a key focus of 4. P. Kopacek, B. Kopacek, Intelligent, flexible disassembly,
future research. The integration of intelligent disassembly Int. J. Adv. Manuf. Technol. 30 (2006) 554–560
technologies has been widely used to significantly improve 5. H. Poschmann, H. Brüggemann, D. Goldmann, Disassembly
the efficient of disassembly process, which assists robotic 4.0: a review on using robotics in disassembly tasks as a way
systems in achieving more intelligent and accurate of automation, Chem. Ing. Tech. 92 (2020) 341–359
disassembly decisions and operations. Additionally, the 6. G. Foo, S. Kara, M. Pagnucco, Challenges of robotic
comprehensive literatures about HRC disassembly tech- disassembly in practice, Proc. CIRP 105 (2022) 513–518
nology has been discussed for the functionality and 7. S. Hjorth, D. Chrysostomou, Human-robot collaboration in
applicability of robot disassembly, which provides the industrial environments: a literature review on non-
diversified solutions for disassembly tasks in different destructive disassembly, Robot. Comput. Integr. Manuf.
complex scenarios. Simultaneously, the comprehensive 73 (2022) 102208
disassembly technologies have been deeply explored to 8. J. Xiao, J. Gao, N. Anwer, B. Eynard, Multi-agent
ensure the integrated safety to reduce the unforeseen risks reinforcement learning method for disassembly sequential
during the disassembly process. By considering HRC task optimization based on human-robot collaborative
disassembly line based on various optimization methods, disassembly in electric vehicle battery recycling, J. Manuf.
digital twin technology in collaborative disassembly can be Sci. Eng. Trans. ASME 145 (2023) 121001
effectively incorporated into HRC disassembly to enhance 9. J. Liu, Z. Xu, H. Xiong, Q. Lin, W. Xu, Z. Zhou, Digital twin-
the efficiency and safety of the disassembly process. driven robotic disassembly sequence dynamic planning
under uncertain missing condition, IEEE Trans. Ind. Inf.
Acknowledgments 19 (2023) 11846–11855
10. A. Keshvarparast, D. Battini, O. Battaia, A. Pirayesh,
The authors are very grateful to all the anonymous reviewers for Collaborative robots in manufacturing and assembly sys-
the valuable opinions and suggestions on the improvement of our tems: literature review and future research agenda, Springer,
paper. USA, 2023
11. J.L. Rickli, J.A. Camelio, Partial disassembly sequencing
Funding considering acquired end-of-life product age distributions,
Int. J. Prod. Res. 52 (2014) 7496–7512
This research is supported by National Natural Science Founda- 12. B. Zhu, U. Roy, Ontology-based disassembly information
tion of China, No. 52305551. system for enhancing disassembly planning and design, Int.
J. Adv. Manuf. Technol. 78 (2015) 1595–1608
13. G. Tian, M.C. Zhou, P. Li, Disassembly sequence planning
Conflicts of interest considering fuzzy component quality and varying opera-
tional cost, IEEE Trans. Autom. Sci. Eng. 15 (2018) 748–
The authors declare that they have no known competing financial 760
interests or personal relationships that could have appeared to 14. Y. Laili, F. Ye, Y. Wang, L. Zhang, Interference probability
influence the work reported in this paper. matrix for disassembly sequence planning under uncertain
interference, J. Manuf. Syst. 60 (2021) 214–225
15. W. Meng, X. Zhang, Optimization of remanufacturing
Data availability statement disassembly line balance considering multiple failures and
material hazards, Sustainable 12 (2020). doi: 10.3390/
Data will be made available on request. SU12187318
 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024) 17

16. M. Liu, X. Liu, F. Chu, F. Zheng, C. Chu, Robust 32. A. Kumar Singh, M.K. Tiwari, S.K. Mukhopadhyay,
disassembly line balancing with ambiguous task processing Modelling and planning of the disassembly processes using
times, Int. J. Prod. Res. 58 (2020) 5806–5835 an enhanced expert Petri net, Int. J. Prod. Res. 41 (2003)
17. X. Zhang et al., Application and planning of an energy- 3761–3792
oriented stochastic disassembly line balancing problem, 33. R. Edmunds, M. Kobayashi, M. Higashi, Using constraint-
Environ. Sci. Pollut. Res. (2023) 0123456789, doi: 10.1007/ satisfaction to optimise disassembly sequences generated from
s11356-023- 27288-4. AND/OR information, Int. J. Prod. Res. 50 (2012) 4105–4126
18. J. Liu et al., Equipment disassembly and maintenance in an 34. S.T. Puente-Mendez, F. Torres-Medina, J. Pomares-Baeza,
uncertain environment based on a peafowl optimization Product disassembly scheduling using graph models,
algorithm, Processes 11 (2023) 1–17 Environmentally Conscious Manufacturing II. SPIE. 4569
19. J. Xiao, C. Jiang, B. Wang, A review on dynamic recycling of (2002) 63–70
electric vehicle battery: disassembly and echelon utilization, 35. M. Turowski, M. Morgan, Y. Tang, Disassembly line design
Batteries 9 (2023) 57 with uncertainty, in 2005 IEEE International Conference on
20. H. Yin, J. Xiao, G. Wang, Human-robot collaboration re- Systems, Man and Cybernetics (2005) pp. 954–959
manufacturing for uncertain disassembly in retired battery 36. M. Turowski, Y. Tang, M. Morgan, Analysis of an adaptive
recycling, in 2022 5th World Conf. Mech. Eng. Intell. Manuf. fuzzy system for disassembly process planning, in 2005 IEEE
WCMEIM 2022 (2022) pp. 595–598, doi: 10.1109/WCME International Symposium on Electronics and the Environ-
IM56910. 2022.10021388 ment, 2005 (2005) pp. 249–254
21. E. Rodriguez, R. Bonnard, A. Alvares, Progresses in the 37. Y. Tang, M.C. Zhou, M. Gao, Fuzzy-petri-net-based
development of a STEP-NC compliant additive manufactur- disassembly planning considering human factors, IEEE
ing system, Produção e Desenvolv. 4 (2018) 39–53 Trans. Syst. Man, Cybern. A 36 (2006) 718–725
22. L. Zhu, Z. Zhang, Y. Wang, N. Cai, On the end-of-life state 38. Y. Tang, A memory fuzzy learning for uncertainty
oriented multi-objective disassembly line balancing prob- management in disassembly, in Proc. 2006 IEEE Int. Conf.
lem, J. Intell. Manuf. 31 (2020) 1403–1428 Networking, Sens. Control. ICNSC’06 (2006) pp. 522–527
23. T. Yu, J. Huang, Q. Chang, Optimizing task scheduling in 39. Y. Tian, T.Y. Wang, B.H. Ding, G.Y. He, H.J. Zheng,
human-robot collaboration with deep multi-agent reinforce- Research on modularization fuzzy petri net of disassembly
ment learning, J. Manuf. Syst. 60 (2021) 487–499 system, in EM (2009) pp. 822–826
24. Y. Laili, F. Tao, D.T. Pham, Y. Wang, L. Zhang, Robotic 40. J.P.B. Santiago Puente Mendez, F. Torres Medina, Product
disassembly re-planning using a two-pointer detection disassembly scheduling using graph models, in Proc. SPIE
strategy and a super-fast bees algorithm, Robot. Comput. 4569, Environmentally Conscious Manufacturing II (2002).
Integr. Manuf. 59 (2019) 130–142 p. 4569
25. M.L. Lee, S. Behdad, X. Liang, M. Zheng, Task allocation and 41. G. Tian, Y. Ren, Y. Feng, M.C. Zhou, H. Zhang, J. Tan,
planning for product disassembly with human-robot collabo- Modeling and planning for dual-objective selective disas-
ration, Robot. Comput. Integr. Manuf. 76 (2022) 102306 sembly using and/or graph and discrete artificial bee colony,
26. J. Liu, Z. Zhou, D.T. Pham, W. Xu, C. Ji, Q. Liu, IEEE Trans. Ind. Inform. 15 (2019) 2456–2468
Collaborative optimization of robotic disassembly sequence 42. B. Zhu, M.I. Sarigecili, U. Roy, Disassembly information model
planning and robotic disassembly line balancing problem incorporating dynamic capabilities for disassembly sequence
using improved discrete Bees algorithm in remanufacturing, generation, Robot. Comput. Integr. Manuf. 29 (2013) 396–409
Robot. Comput. Integr. Manuf. 61 (2020) 101829 43. F. Ye, J. Perrett, L. Zhang, Y. Laili, Y. Wang, A self-
27. X.F. Zhang, S.Y. Zhang, Product cooperative disassembly evolving system for robotic disassembly sequence planning
sequence planning based on branch-and-bound algorithm, under uncertain interference conditions, Robot. Comput.
Int. J. Adv. Manuf. Technol. 51 (2010) 1139–1147 Integr. Manuf. 78 (2022) 102392
28. J. Liu, Z. Zhou, D. T. Pham, W. Xu, C. Ji, Q. Liu, 44. K. Meng, G. Xu, X. Peng, K. Youcef-Toumi, J. Li, Intelligent
Collaborative optimization of robotic disassembly sequence disassembly of electric-vehicle batteries: a forward-looking
planning and robotic disassembly line balancing problem overview, Resour. Conserv. Recycl. 182 (2022) 106207
using improved discrete Bees algorithm in remanufacturing, 45. A. Billard, D. Kragic, Trends and challenges in robot
Robot. Comput. Integr. Manuf. 61 (2020) 101829 manipulation, Science (80-) 364 (2019) eaat8414
29. J. Gao, G. Wang, J. Xiao, P. Zheng, E. Pei, Partially 46. G. Caiza, C.A. Garcia, J.E. Naranjo, M.V. Garcia, Flexible
observable deep reinforcement learning for multi-agent robotic teleoperation architecture for intelligent oil fields,
strategy optimization of human-robot collaborative disas- Heliyon 6 (2020) e03833
sembly: a case of retired electric vehicle battery, Robot. 47. A. Weigl, Requirements for robot assisted disassembly of not
Comput. Integr. Manuf. 89 (2024) 102775 appropriately designed electronic products: lessons from first
30. Y. Yin, P. Zheng, C. Li, L. Wang, A state-of-the-art survey studies, in 1994 IEEE International Symposium on
on augmented reality-assisted digital twin for futuristic Electronics and The Environment (1994) pp. 337–342
human-centric industry transformation, Robot. Comput. 48. K. Hohm, H.M. Hofstede, H. Tolle, Robot assisted
Integr. Manuf. 81 (2023) 1–21 disassembly of electronic devices, IEEE Int. Conf. Intell.
31. J. Xiao, N. Anwer, W. Li, B. Eynard, C. Zheng, Dynamic Robot. Syst. 2 (2000) 1273–1278
Bayesian network-based disassembly sequencing optimiza- 49. K. Wegener, W.H. Chen, F. Dietrich, K. Dröder, S. Kara,
tion for electric vehicle battery, CIRP J. Manuf. Sci. Robot assisted disassembly for the recycling of electric
Technol. 38 (2022) 824–835 vehicle batteries, Procedia CIRP 29 (2015) 716–721
18 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024)

50. T. Tolio et al., Design, management and control of 69. J.P.J. Prioli, J.L. Rickli, Collaborative robot based
demanufacturing and remanufacturing systems, CIRP architecture to train flexible automated disassembly systems
Ann.  Manuf. Technol. 66 (2017) 585–609 for critical materials, Proc. Manuf. 51 (2020) 46–53
51. R. Bogue, Robots in recycling and disassembly, Ind. Robot 70. J. Zhou, Y. Zhou, B. Wang, J. Zang, Human-cyber-physical
Int. J. Robot. Res. Appl. 46 (2019) 461–466 systems (HCPSs) in the context of new-generation intelli-
52. A. Ajoudani, A.M. Zanchettin, S. Ivaldi, A. Albu-Schäffer, K. gent manufacturing, Engineering 5 (2019) 624–636
Kosuge, O. Khatib, Progress and prospects of the human- 71. S. Lou, R. Tan, Y. Zhang, C. Lv, Human-robot interactive
robot collaboration, Auton. Robots 42 (2018) 957–975 disassembly planning in Industry 5.0, IEEE/ASME Int.
53. Q. Liu, Z. Liu, W. Xu, Q. Tang, Z. Zhou, D. Truong, Human- Conf. Adv. Intell. Mechatronics, AIM (2023) pp. 891–895
robot collaboration in disassembly for sustainable manufactur- 72. C.S.A. Assuad, T. Leirmo, K. Martinsen, Proposed
ing, Int. J. Prod. Res. 57 (2019) 4027–4044 framework for flexible de- and remanufacturing systems
54. Z. Liu, Deep learning-based human motion prediction using cyber-physical systems, additive manufacturing, and
considering context awareness for human-robot collabora- digital twins, Proc. CIRP 112 (2022) 226–231
tion in manufacturing, Proc. CIRP 83 (2019) 272–278 73. N. Nikolakis, A cyber physical system (CPS) approach for
55. M.L. Lee, S. Behdad, X. Liang, M. Zheng, Disassembly safe human-robot collaboration in a shared workplace,
sequence planning considering human-robot collaboration, Robot. Comput. Integr. Manuf. 56 (2019) 233–243
Proc. Am. Control Conf. 2020-July, no. 5 (2020) pp. 2438– 74. N. Almasarwah, E. Abdelall, M. Saraireh, S. Ramadan,
2443 Human-Cyber-Physical Disassembly Workstation 5.0 for
56. M.L. Lee, S. Behdad, X. Liang, M. Zheng, A real-time Sustainable Manufacturing, 2022 Int. Conf. Emerg. Trends
receding horizon sequence planner for disassembly in a Comput. Eng. Appl. ETCEA 2022  Proc. (2022) pp. 1–5
human-robot collaboration setting, in 2020 Int. Symp. Flex. 75. G. Ferrer, S.K. Heath, N. Dew, An RFID application in large
Autom. ISFA 2020 (2020), doi: 10.1115/ISFA2020-9657 job shop remanufacturing operations, Int. J. Prod. Econ.
57. M.L. Lee, W. Liu, S. Behdad, X. Liang, M. Zheng, Robot- 133 (2011) 612–621
assisted disassembly sequence planning with real-time 76. M.M.L. Chang, Approaches and challenges in product
human motion prediction, IEEE Trans. Syst. Man. Cybern. disassembly planning for sustainability, Proc. CIRP 60
Syst. 53 (2023) 438–450 (2017) 506–511
58. Y. Ding, W. Xu, Z. Liu, Z. Zhou, D.T. Pham, Robotic task 77. J. Wessel, A. Turetskyy, O. Wojahn, C. Herrmann, S.
oriented knowledge graph for human-robot collaboration in Thiede, Tracking and tracing for data mining application in
disassembly, Proc. CIRP 83 (2019) 105–110 the lithium-ion battery production, Proc. CIRP 93 (2020)
59. S. Tian, X. Liang, M. Zheng, An optimization-based human 162–167
behavior modeling and prediction for human-robot collabo- 78. A.K. Gulivindala, M.V.A.R. Bahubalendruni, R. Chandra-
rative disassembly, Proc. Am. Control Conf. (2023) pp. sekar, E. Ahmed, M.H. Abidi, A. Al-Ahmari, Automated
3356–3361 disassembly sequence prediction for industry 4.0 using
60. W. Xu, Q. Tang, J. Liu, Z. Liu, Z. Zhou, D.T. Pham, enhanced genetic algorithm, Comput. Mater. Contin. 69
Disassembly sequence planning using discrete Bees algo- (2021) 2531–2548
rithm for human-robot collaboration in remanufacturing, 79. P.K.R. Maddikunta, G. Srivastava, T.R. Gadekallu, N.
Robot. Comput. Integr. Manuf. 62 (2020) 101860 Deepa, P. Boopathy, Predictive model for battery life in IoT
61. X. Guo et al., Human-robot collaborative disassembly line networks, IET Intell. Transp. Syst. 14 (2020) 1388–1395
balancing problem with stochastic operation time and a 80. A.D. Joshi, S.M. Gupta, Evaluation of design alternatives of
solution via multi-objective shuffled frog leaping algorithm, End-Of-Life products using internet of things, Int. J. Prod.
IEEE Trans. Autom. Sci. Eng. (2023) 1–12 Econ. 208 (2019) 281–293
62. G. Wang, J. Gao, J. Xiao, Optimal strategy of disassembly 81. R. Sommerville, J. Shaw-Stewart, V. Goodship, N. Rowson,
process in electric vehicle battery based on human-machine E. Kendrick, A review of physical processes used in the safe
collaboration re-manufacturing, 2023 9th Int. Conf. Mecha- recycling of lithium ion batteries, Sustain. Mater. Technol.
tronics Robot. Eng. ICMRE 2023 (2023) pp. 174–179 25 (2020) e00197
63. E.A. Lee, Cyber physical systems: design challenges (2008) 82. O. Ondemir, S.M. Gupta, Quality management in product
64. P. Kopacek, Trends in production automation, IFAC- recovery using the Internet of Things: an optimization
PapersOnLine 52 (2019) 509–512 approach, Comput. Ind. 65 (2014) 491–504
65. R. Alguliyev, Y. Imamverdiyev, L. Sukhostat, Cyber- 83. H.-C. Yi, J.W. Park, Design and implementation of an end-
physical systems and their security issues, Comput. Ind. of-life vehicle recycling center based on IoT (Internet of
100 (2018) 212–223 Things) in Korea, Proc. CIRP 29 (2015) 728–733
66. A. Hahn, R.K. Thomas, I. Lozano, A. Cardenas, A multi- 84. Y. Zhang et al., The ‘Internet of Things’ enabled real-time
layered and kill-chain based security analysis framework for scheduling for remanufacturing of automobile engines, J.
cyber-physical systems, Int. J. Crit. Infrastruct. Prot. 11 Clean. Prod. 185 (2018) 562–575
(2015) 39–50 85. K. Xia, L. Gao, W. Li, L. Wang, K.-M. Chao, A Q-learning
67. N. Nikolakis, K. Sipsas, S. Makris, A cyber-physical context- based selective disassembly planning service in the cloud
aware system for coordinating human-robot collaboration, based remanufacturing system for WEEE (2014), doi:
Proc. CIRP 72 (2018) 27–32 10.1115/MSEC2014-4008
68. A. Filipescu et al., Modeling and control an A/DT served by an 86. H. Jiang, J. Yi, S. Chen, X. Zhu, A multi-objective algorithm
ACPS based on SCADA, Industry 4.0 and 5.0, IEEE Int. Conf. for task scheduling and resource allocation in cloud-based
Emerg. Technol. Fact. Autom. ETFA 2023 (2023) pp. 1–8 disassembly, J. Manuf. Syst. 41 (2016) 239–255
 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024) 19

87. F. Mancusi, A cloud-aided remanufacturing framework to 103. F. Tao, L. Bi, Y. Zuo, A.Y.C. Nee, Partial/parallel
assess the relative complexity, IFAC-PapersOnLine 55 disassembly sequence planning for complex products, J.
(2022) 1025–1030 Manuf. Sci. Eng. Trans. ASME 140 (2018). doi: 10.1115/
88. T.F. Go, D.A. Wahab, M.N.A. Rahman, R. Ramli, A. 1.4037608
Hussain, Genetically optimised disassembly sequence for 104. C.B. Kalayci, S.M. Gupta, A tabu search algorithm for
automotive component reuse, Expert Syst. Appl. 39 (2012) balancing a sequence-dependent disassembly line, Prod.
5409–5417 Plan. Control 25 (2014) 149–160
89. M. Kheder, M. Trigui, N. Aifaoui, Disassembly sequence 105. W. Liang, Z. Zhang, T. Yin, Y. Zhang, T. Wu, Modelling and
planning based on a genetic algorithm, Proc. Inst. Mech. optimisation of energy consumption and profit-oriented
Eng. Part C J. Mech. Eng. Sci. 229 (2015) 2281–2290 multi-parallel partial disassembly line balancing problem,
90. H.-E. Tseng, S.-C. Lee, Disassembly sequence planning Int. J. Prod. Econ . 262 (2023) 108928
using interactive genetic algorithms, ICNC-FSKD 2018- 106. X. Xia, W. Liu, Z. Zhang, L. Wang, J. Cao, X. Liu, A
14th Int. Conf. Nat. Comput. Fuzzy Syst. Knowl. Discov. balancing method of mixed-model disassembly line in
(2018) pp. 77–84 random working environment, Sustain 11 (2019), doi:
91. C. Lu, H.Z. Huang, B. Zheng, J.Y.H. Fuh, Y.S. Wong, An 10.3390/su11082304
Ant colony optimization approach to disassembly planning, 107. Y. Fang, H. Ming, M. Li, Q. Liu, D.T. Pham, Multi-objective
in 2008 Int. Conf. Apperceiving Comput. Intell. Anal. evolutionary simulated annealing optimisation for
ICACIA 2008 (2008) pp. 81–84 mixed-model multi-robotic disassembly line balancing
92. S.M. McGovern S.M. Gupta, Ant colony optimization for with interval processing time, Int. J. Prod. Res. 58 (2020)
disassembly sequencing with multiple objectives, Int. J. 846–862
Adv. Manuf. Technol. 30 (2006) 481–496 108. K. Wang, X. Li, L. Gao, P. Li, S.M. Gupta, A genetic
93. B. Hu, Y. Feng, H. Zheng, J. Tan, Sequence planning for simulated annealing algorithm for parallel partial disassem-
selective disassembly aiming at reducing energy consump- bly line balancing problem, Appl. Soft Comput. 107 (2021)
tion using a constraints relation graph and improved ant 107404
colony optimization algorithm, Energies 11 (2018). doi: 109. M. Chu, W. Chen, Human-robot collaboration disassembly
10.3390/en11082106 planning for end-of-life power batteries, J. Manuf. Syst. 69
94. J. Liu, Z. Zhou, D.T. Pham, W. Xu, C. Ji, Q. Liu, Robotic (2023) 271–291
disassembly sequence planning using enhanced discrete bees 110. A. Allagui, I. Belhadj, R. Plateaux, M. Hammadi, O. Penas,
algorithm in remanufacturing, Int. J. Prod. Res. 56 (2018) N. Aifaoui, Reinforcement learning for disassembly sequence
3134–3151 planning optimization, Comput. Ind. 151 (2023). doi:
95. N. Hartono, F. Javier Ramírez, D.T. Pham, A sustainabili- 10.1016/j.compind.2023.103992
ty-based model for robotic disassembly sequence planning in 111. E. Yildiz, F. Worgotter, DCNN-based screw detection for
remanufacturing using the Bees Algorithm, IFAC-Paper- automated disassembly processes, in Proc.  15th Int. Conf.
sOnLine 55 (2022) 1013–1018 Signal Image Technol. Internet Based Syst. SISITS 2019
96. C.B. Kalayci, S.M. Gupta, Artificial bee colony algorithm (2019) pp. 187–192
for solving sequence-dependent disassembly line balancing 112. X. Li, M. Li, Y. Wu, D. Zhou, T. Liu, F. Hao, Accurate screw
problem, Expert Syst. Appl. 40 (2013) 7231–7241 detection method based on faster R-CNN and rotation edge
97. W. Zhang, Y. Zheng, R. Ahmad, The integrated process similarity for automatic screw disassembly, Int. J. Comput.
planning and scheduling of flexible job-shop-type remanu- Integr. Manuf. 34 (2021) 1177–1195
facturing systems using improved artificial bee colony 113. M.F. Adesso, R. Hegewald, N. Wolpert, E. Schomer, B.
algorithm, J. Intell. Manuf. 34 (2023) 2963–2988 Maier, B.A. Epple, Automatic classification and disassembly
98. C. Pornsing, A. Watanasungsuit, Discrete particle swarm of fasteners in industrial 3D CAD-scenarios, in 2022 IEEE
optimization for disassembly sequence planning, International Conference on Robotics and Automation
ICMIT2014-2014 IEEE Int. Conf. Manag. Innov. Technol. (ICRA) (2022) pp. 9874–9880
(2014) pp. 480–485 114. D.P. Brogan, N.M. DiFilippo, M.K. Jouaneh, Deep learning
99. H.-E. Tseng, C.-C. Chang, T.-W. Chung, Applying improved computer vision for robotic disassembly and servicing
particle swarm optimization to asynchronous parallel disas- applications, Array 12 (2021) 100094
sembly planning, IEEE Access 10 (2022) 80555–80564 115. S. Mangold, C. Steiner, M. Friedmann, J. Fleischer, Vision-
100. Y. Tao, K. Meng, P. Lou, X. Peng, X. Qian, Joint decision- based screw head detection for automated disassembly for
making on automated disassembly system scheme selection remanufacturing, Proc. CIRP 105 (2022) 1–6
and recovery route assignment using multi-objective meta- 116. X. Zhang, K. Eltouny, X. Liang, S. Behdad, Automatic
heuristic algorithm, Int. J. Prod. Res. 57 (2019) 124–142 screw detection and tool recommendation system for robotic
101. C.B. Kalayci, S.M. Gupta, A particle swarm optimization disassembly, J. Manuf. Sci. Eng. 145 (2023) 1–8
algorithm with neighborhood-based mutation for sequence- 117. X. Zhang, D. Yi, S. Behdad, S. Saxena, Unsupervised human
dependent disassembly line balancing problem, Int. J. Adv. activity recognition learning for disassembly tasks, IEEE
Manuf. Technol. 69 (2013) 197–209 Trans. Ind. Informatics 20 (2024) 785–794
102. M. Alshibli, A. El Sayed, E. Kongar, T.M. Sobh, S.M. 118. W. Deng et al., Predictive exposure control for vision-based
Gupta, Disassembly sequencing using tabu search, J. Intell. robotic disassembly using deep learning and predictive
Robot. Syst. Theory Appl. 82 (2016) 69–79 learning, Robot. Comput. Integr. Manuf. 85 (2024) 102619
20 J. Xiao and K. Huang: Manufacturing Rev. 11, 17 (2024)

119. Y. Chen, H.Y. Liao, S. Behdad, B. Hu, Human activity 126. M. Kerin, D.T. Pham, J. Huang, J. Hadall, A generic asset
recognition in an end-of-life consumer electronics disassem- model for implementing product digital twins in smart
bly task, Appl. Ergon. 113 (2023) 104090 remanufacturing, Int. J. Adv. Manuf. Technol. 124 (2023)
120. F. Ranz, V. Hummel, W. Sihn, Capability-based task 3021–3038
allocation in human-robot collaboration, Proc. Manuf. 9 127. C. Mouflih, R. Gaha, A. Durupt, M. Bosch-Mauchand, K.
(2017) 182–189 Martinsen, B. Eynard, Decision support framework using
121. F. Chen, K. Sekiyama, F. Cannella, T. Fukuda, Optimal knowledge based digital twin for sustainable product
subtask allocation for human and robot collaboration within development and end of life, Proc. Des. Soc. 3 (2023)
hybrid assembly system, IEEE Trans. Autom. Sci. Eng. 11 1157–1166
(2014) 1065–1075 128. W. Qu, J. Li, R. Zhang, S. Liu, J. Bao, Adaptive planning of
122. B. Wu, B. Hu, H. Lin, Toward efficient manufacturing human-robot collaborative disassembly for end-of-life lithi-
systems: a trust based human robot collaboration, Proc. Am. um-ion batteries based on digital twin, J. Intell. Manuf.
Control Conf. (2017) 1536–1541 (2023). doi: 10.1007/s10845-023- 02081-9
123. C. Mouflih, R. Gaha, M. Bosch, A. Durupt, Improving the 129. J. Li, H. Zheng, J. Bao, Development of digital twin-based
Sustainability of Manufactured Products: Literature Review WEEE recycling workshop: initial approaches and future
and Challenges for Digitalization of Disassembly and challenges, in 2022 6th Int. Conf. Robot. Autom. Sci. ICRAS
Dismantling Processes 667 IFIP (Springer Nature 2022 (2022) pp. 231–237
Switzerland, 2023) 130. W. Li, Y. Peng, Y. Zhu, D.T. Pham, A.Y.C. Nee, S.K. Ong,
124. A. Baratta, A. Cimino, F. Longo, L. Nicoletti, Computers & End-of-life electric vehicle battery disassembly enabled by
industrial engineering digital twin for human-robot collabo- intelligent and human-robot collaboration technologies: a
ration enhancement in manufacturing systems: literature review, Robot. Comput. Integr. Manuf. 89 (2024) 102758
review and direction for future developments, Comput. Ind. 131. J. Gao, G. Wang, J. Xiao, P. Zheng, E. Pei, Partially observable
Eng. 187 (2024) 109764 deep reinforcement learning for multi-agent strategy optimi-
125. E. Verna, S. Puttero, G. Genta, M. Galetto, Toward a zation of human-robot collaborative disassembly: a case of
concept of digital twin for monitoring assembly and retired electric vehicle battery, Robot. Comput. Integr. Manuf.
disassembly processes, Qual. Eng. 0 (2023) 1–18 89 (2024). doi: 10.1016/j.rcim.2024.102775

Cite this article as: Jinhua Xiao, Kaile Huang, A comprehensive review on human–robot collaboration remanufacturing towards
uncertain and dynamic disassembly, Manufacturing Rev. 11, 17 (2024)