Smart and Sustainable Human-Centred Workstations for Operators with Disability in the Age of Industry 5.0: A Systematic Review

Abstract

The World Bank has reported that over one billion individuals have a disability, implying that almost fifteen percent of the global inhabitants are susceptible to undergoing levels of discrimination, especially in employment. This issue may prevail on a manufacturing shop floor, whereby a wave of standardisation dominates such as in the design of shop floor workstations. Despite advances made in the literature, people with disabilities are still siloed from manufacturing. Consequently, the aim of this research work was to analyse literature’s current state of the art on the design of workstations for operators with disabilities within the context of Industry 5.0, where sustainability, human-centricity, and resilience are upheld. The study employed a systematic review of 69 publications from Scopus and Google Scholar published between 2013 and 2023, adhering to the updated PRISMA guidelines to identify the major research gaps. The review contributes an understanding of the current academic and industrial limitations such as the absence of social applicability of Industry 4.0 technology, the rift between academic knowhow and industrial implementation, and the lack of alignment with the sustainable development goals (SDGs). Additionally, the review uncovered an absence in work bridging four disciplines together: workstation design, Industry 5.0, sustainability, and disability. An unprecedented understanding of the interdependency between all four disciplines within the remit of smart, sustainable, and inclusive manufacturing workstations is contributed. This review proposes directions amidst the four most relevant SDGs—SDGs 8, 9, 10, and 12 to the topic.

1. Introduction

The fourth industrial revolution, also referred to as Industry 4.0, aimed to accommodate the evolution of digital and interconnected technologies and stemmed from the previous three industrial revolutions. Via the implementation of artificial intelligence, IoT, cyber-security, and big data analytics, manufacturing is giving way to connected machinery, flexibility, and personalised production [1,2]. However, despite the higher level of productivity and efficiency, such advances may take a toll on the operators’ skills and well-being [3,4]. Today’s shop floor consists of a pool of workers, with ranging discrepancies in ages, abilities, training, and experience [5,6,7,8], thus launching a domino effect on various production factors such as cost, time, health, and safety [5]. Consequently, such diversification and the need to safeguard the human worker has resulted in the development of the Industry 5.0 ideology. Within this concept, a triad is established between sustainability (social, economic, and environmental), human-centricity, and resilience within manufacturing [9,10]. Industry 5.0 is geared at prioritising the needs of the operator on multiple levels, in an attempt to harmonise human-centricity with the currently established techno-centricity. Such attention to the operator, however, should not solely be funnelled towards mainstream operators, but should equally target minorities. Therefore, the prospect of Industry 5.0 broadens the horizon for diversity and inclusion [11,12], especially for people with disabilities who may be estranged from the manufacturing shop floor due to an absence of accessibility, stigma, and other lingering reasons. The United Nations (UN) [13] explain the term “disability” as “long-term physical, mental, intellectual or sensory impairments”. This definition shall thus be adopted throughout this research work.

Before understanding why Industry 5.0 and inclusion should work together in manufacturing, it is worth understanding the challenges faced by persons with disabilities, and how these may impact their employment [5]. Persons with disabilities may face prejudice, employment discrimination, inferior education and lack of access to information, higher risk of redundancy, and scant probability of re-recruitment and training, all leading to a higher risk of poverty [4,14,15]. Such an observation leaves an ingrained effect, especially when considering that 87 million Europeans are persons with disability, just of which half (50%) are employed [16]. With the prospect of Industry 5.0, the manufacturing sector would be able to broaden employment possibilities to more operators with a disability. This potential is far from futile, notably when recalling that in 2020 alone, more than 29 million people were employed within the EU manufacturing sector [17]. This aptly justifies the importance given to human-centric manufacturing in Industry 5.0. In the state of art, there already is a plethora of literature surveys that tackle certain facets of the problem. Xu et al. [9] elucidate the perspective of shifting from Industry 4.0 to Industry 5.0, and the implications thereof. In line with Industry 5.0’s main areas of priority, Katiraee et al. [5] propounded a literature survey on human-factor considerations in manufacturing. Another literature survey was conducted by Grybauskas et al. [12], harmonising the social aspect of sustainability with the effects yielded by the techno-centric Industry 4.0. That being said, none of these scholarly works take a unified approach which considers the role of an operator with disabilities in a manufacturing company that operates under the auspices of Industry 5.0’s focus on human-centricity and sustainability.

Mark et al. establish that there is a shortcoming of practical scenarios that are specific to people with disabilities on the manufacturing shop floor, and that the absence of incentives at industry and corporate level heighten these hurdles [18]. One of the persisting challenges in academia is that disciplines such as ergonomics, human–robot collaboration, and workstation design narrow down on the needs of “able-bodied workers” [19] (p. 1117). Standardisation (such as adopting touch screens for everybody [20]) is oblivious to the notion that “one size does not fit all” and tramples on comprehending how different abilities necessitate alternative design approaches, technology, and task assignments. On the other hand, Bento and Kuznetsova [21] showcase how certain employers in Norway view persons with disabilities. Some employers commented that they felt it difficult to adapt their workstations to be inclusive, claiming that their company was “not a kindergarten [that has] to arrange the work tasks specifically for these people” [21] (p. 38). Such statements reiterate the fact that most employers within the manufacturing industry are deterred from employing persons with disabilities due to the perception that adapting workstations for persons with disabilities is a challenging feat and would require much more assistance than a mainstream worker (even if the task is familiar) [21,22,23]. This reflects the lack of design guidelines and knowledge, and it aids at the disposition of employers that would help achieve this goal [11,23]. On the contrary, however, Litwin et al. [24] elaborate on the benefits observed when employing people with diverse abilities on the shop floor, provided that the necessary amendments are heeded to. The authors also acknowledge potential negative implications, such as additional workforce limitations and the requirement of further planning. Consequently, Romero et al. [25] advocate the ability of “adaptive physical and cognitive automation” [25] (p. 9) to grant operators with disabilities the capability to carry out the same tasks as mainstream operators with ease.

Disability inclusion also comprises a fundamental goal in the 2030 Sustainable Development Agenda [26], which claims that no person should be “left behind” as a result of their disability, such as when seeking employment. The UN furthers this promotion of sustainability through a list of seventeen global goals, known as the Sustainable Development Goals (SDGs). According to the UN, the SDGs that address persons with disabilities are SDG 4, 8, 10, 11, and 17 [27]. It is important to add that SDGs governing persons with disabilities should ensure that this social and inclusive aspect of sustainability within manufacturing is not merely considered, but also targeted. With the ubiquity of Industry 5.0, fast-paced manufacturing, and high-end technologies, it thus becomes crucial that all three counterparts (technology, inclusion, and sustainability) work in unison on the onset of design without compromising on either. Subsequently, this research work employs the following research question to capture the aforementioned areas:

What is the current state of the art regarding the design of inclusive smart and sustainable workstations for operators with disabilities on the manufacturing shop floor?

This question can be viewed as an amalgamation of four main research disciplines which will scaffold the review:

• Discipline 1: Workstation Design.
• Discipline 2: Disability in Manufacturing.
• Discipline 3: Industry 4.0 and Industry 5.0.
• Discipline 4: Sustainability and Social Sustainability.

The golden thread between disability, sustainability, workstation design, and manufacturing may initially present itself as obscure and blurred, thus warranting in-depth research to better comprehend this research niche and its ensuing interdependencies. The human-centric prospect of Industry 5.0 ties seamlessly with the lesser explored social facet of sustainability, especially when the spotlight is on inclusive manufacturing workstation design. Industry 5.0 should, however, reap the benefits established through Industry 4.0 by embracing enabling smart manufacturing technologies to provide workstations that exclude no one on the shopfloor.

The objective of this study is therefore to provide a concrete review of the state of the art as denoted in this proposed outline, whilst attempting to intertwine the four disciplines together. Section 2 introduces the PRISMA approach and the corresponding search terms, followed by a thorough statistical analysis of the literature in Section 3. Section 4 delves into current workstation design for operators with disabilities on the manufacturing shop floor and provides the main take-aways for practitioners and academics (Disciplines 1 and 2). Similarly, Section 5 and Section 6 investigate the role of Industry 5.0 and sustainability, respectively, vis-à-vis inclusivity on the shop floor (Disciplines 3 and 4). Finally, Section 7 attempts to answer the research question based on the results of the literature review by providing a snapshot of the review’s outcomes along with the contribution of this study towards sustainable manufacturing and potential for fruitful future work in the domain of smart, sustainable, and inclusive workstations.

2. Setting the Scene for the PRISMA Systematic Review

Based on the initial research carried out and the above-mentioned argumentation, the search terms exhibited in

Table 1. A list of search terms that were pivotal to conduct this systematic review.

“inclusive” AND “manufacturing” AND “workstation” AND “Industry 4.0” AND “disabilit”

“industry 4.0” AND “disabilit” AND “workstation” AND “manufacturing”

“smart” AND “sustainab” AND “workstation” AND “manufacturing” AND “disabilit”

“sustainab” AND “workstation” AND “industry 4.0” AND “disabilit”

“sustainab” AND “workstation” AND “industry 5.0” AND “disabilit”

“industry 5.0” AND “disabilit” AND “manufacturing”

“industry 4.0” AND “sustainab” AND “manufacturing”

“industry 5.0” AND “sustainab” AND “manufacturing”

were selected for this review, with keywords from the four disciplines strung together using the “AND” operator to converge the searches. Scopus was used as the main database during the commencement of the literature search, despite additional publications from other databases such as Google Scholar being used in congruence. The latter publications were eventually added to the review after being deemed relevant for answering the two posed research questions.

Applying the PRISMA method for identifying appropriate studies for this review, the search window was constricted to 69 results, as shown in Figure 1. This pool of publications encompasses journal articles and conference papers spread over the decade between 2013 and 2023. A decade was deemed sufficient to monitor the changes of a rapidly evolving sector such as that of manufacturing and Industry 4.0/Industry 5.0.

3. Statistical Detail following the Identification of the Literature

This section of the review paper involves key statistical and logistical data extracted from the eligible body of literature in order to provide context and allow for the reader’s judgement throughout. Primarily, Figure 2 highlights the number of journal articles and conference papers over the span of a decade (2013–2023). It can be clearly observed that scholarly work associated with two or more of the four disciplines identified earlier is still in its embryonic stage (peaking in the previous year, 2022), mirroring the infantile stages of Industry 5.0 (as noted by [29]) and paving the way for more research work in human-centred and sustainable manufacturing.

Figure 3 presents a VOSViewer Author Density Visualisation Map to capture the above.

The clustered way by which the authors are shown implies a stark segregation between authors from different disciplines. A few authors focus solely on Industry 4.0, human factors in manufacturing, and Operator 4.0 (such as E. Rauch and D. Romero), whereas the clusters from disability studies are still quite narrow and with minimal to no co-concurrence with other authors. This amplifies the possibility of further exploring interdisciplinary work in this domain. Such an observation can be seconded through the interpretation of Figure A1 and Figure A2 in Appendix A, which present pie-charts for the journal publications and conference titles, respectively. The most popular journal was “Computers and Industrial Engineering”; however other prominent journals from the humanities sphere were also identified (such as “Life Span and Disability” and “Research in Developmental Disabilities”). Similarly, the most popular conference was “PETRA—PErvasive Technologies Related to Assistive Environments Conference”, which bridges two of the four introduced disciplines (disability and Industry 4.0). The presence of diverse journals and conferences (not just one or two governing ones) highlights potential for even more studies that merge these four disciplines into one, as is also evident through the list of the top ten distinct cited papers in

Table 2. A list of the 10 most-cited publications reviewed for this literature review.

Rank	Publication Title	Author	Date
1st	Industry 4.0, digitization, and opportunities for sustainability	M. Ghobakhloo [31]	2020
2nd	Industry 4.0 technologies assessment: A sustainability perspective	C. Bai [32]	2020
3rd	Industry 5.0: A survey on enabling technologies and potential applications	P. Maddikunta [10]	2022
4th	Industry 4.0 and Industry 5.0—Inception, conception and perception	X. Xu [9]	2021
5th	Empowering and engaging industrial workers with Operator 4.0 solutions	E. Kaasinen [2]	2020
6th	Working with Augmented Reality?: A Long-Term Analysis of In-Situ Instructions at the Assembly Workplace	M. Funk [33]	2017
7th	The role of manufacturing in affecting the social dimension of sustainability	J. Sutherland [34]	2016
8th	The Design Space of Augmented and Virtual Reality Applications for Assistive Environments in Manufacturing: A Visual Approach	S. Büttner [35]	2017
9th	Towards a Human-Centred Reference Architecture for Next Generation Balanced Automation Systems: Human-Automation Symbiosis	D. Romero [25]	2015
10th	Using In-Situ Projection to Support Cognitively Impaired Workers at the Workplace	M. Funk [36]	2015

.

It is worth noting how the search terms related to “Industry 5.0” AND “manufacturing”, which is a relatively new concept, has amassed more than half of the top ten cited papers within this review. Hence, since the results are indicative of the weight and interest that the novel Industry 5.0 has gained, this anchors the study‘s complementing of Industry 5.0’s principles even more. Another significant insight is that only two (Paper 8 and Paper 10) of the top ten cited publications exhibited in Table 2 allude in some way to Discipline 2 (i.e., Disability in Manufacturing), mainly focusing on assistance provision. This observation alone is already sufficient to compel further interdisciplinary work that steers away from just the mainstream operator and envelopes inclusivity on the shop floor.

Furthermore, Figure 4 provides an even more comprehensive insight into the weight and correlation between the most important keywords elicited through the PRISMA review. Once again, “Industry 4.0” has close knit ties with other heavy-weight terms such as “ergonomics”, “assembly”, and “design”, but also with keywords with less frequency such as “social sustainability” and “sustainable development”. This visualisation map affirms that a relationship is indeed established between “Workstation Design”, “Industry 4.0 and Industry 5.0”, and “Sustainability and Social Sustainability” (three out of four disciplines). When it comes to Discipline 2 (i.e., Disability in Manufacturing), keywords such “operators with disability/ies”, “inclusive manufacturing”, or “accessible manufacturing” are still quite restrained and do not fall under the highest mentions, despite the reviewed publications being oriented around and prioritising this discipline in particular. This clearly suggests that although the links between Industry 4.0, sustainability, and workstation design are quite strong and avid (three out of four disciplines), an insight into relating the three disciplines with the fourth (disability) is still absent due to scant work done in this regard. To summarise, despite addressing people with disabilities on workstations, there is not one study which coalesces the four disciplines into unity.

Following a briefing on the quantitative detail surrounding the body of work chosen through the PRISMA, the upcoming Section 4, Section 5 and Section 6 provide a comprehensive review on the main themes surrounding the body of work selected. Subsequently, Section 7 highlights emergent research gaps and sets the scene for future work to be undertaken in the remit of smart, sustainable, and inclusive workstation design for manufacturing shop floors. The same section simultaneously provides a constructive summary of whether the research question introduced in Section 1 was answered, as well as the contribution of this systematic review to knowledge.

4. Workstation Design for Heterogenous Operators on the Manufacturing Shop Floor

Appropriate task scheduling calls for flexible processes, optimal resource deployment, instantaneous response to erroneous processes or bottlenecks, and a reduction in downtime [2]. One may revert to deconstructing a mere assembly task into sub-tasks, which may be categorised as routine operations (carried out by medium skilled operators), abstract operations (necessitating a high degree of creativity and skill), and manual operations (mainly delegated to operators with a lower level of skill) [6]. Despite each manufacturing company fostering its preference towards assembly line design and work assignments, such procedures may need to be revised for operators with disabilities, owing to performance discrepancies [37] and, hence, ambiguities on how to reach optimal efficiency [24].

4.1. Designing Workstations That Empower Operators with Disabilities as Smart Learners

Conveying information becomes a task which calls for even more focus. A fundamental element often overlooked on the manufacturing shop floor is that the workstation operator is ultimately also a continuous learner [29,38,39], and thus the way information is transmitted and received plays a crucial role in forming learning patterns [7]. Indeed, many researchers have followed the concept of Operator 4.0 and have upheld the latter as a scaffold for operator empowerment within Industry 5.0. The Operator 4.0 concept thrives on exploiting technology to enhance one’s best potential through the training [2,10] and provision of ample practical learning skills. Smart factories are intended to be equipped with operators possessing individual preferences and attributes, and it is thus vital that the shop floor is apt for these diverse portfolios of skills and abilities [2], whilst concurrently prompting the operator to flexibly reach their full potential with the technological assistance needed. Focus is lent to how a smart system should adapt to the abilities of its human counterpart: physical and ergonomic conformance to the operator; adherence to the skills portfolio of the operator; consideration of the preferred mode of interaction; and adaptation based on the level of automation (termed LoA), learning, and training patterns [2,7,11]. This is corroborated by Villani et al., who dedicated one of three models (“Measure”, “Adapt” and “Teach”) within their proposed framework towards adaptability, whereby tasks are customised to cater for the operator’s ability portfolio [40]. Desideri et al. [41] bring light to the prospectives that operators with disabilities would like to achieve, namely, the ability to follow sequential tasks autonomously. Having established that, the authors do not shy away from mentioning the requirements for successful completion of step-by-step tasks by individuals with intellectual and development disabilities (IDDs). Primarily, the user must be able to differentiate between each successive step and then apply the chosen method of control to navigate within, say, the graphical user interface. Despite these two obligations being taken for granted by designers and mainstream operators, the authors note how for people with severe IDDs, this may be a hurdle to overcome, and thus technology should step in to compensate. This sentiment is argued further in Section 4.2, which details the prospects of holistically inclusive workstation design (for, with, and by) and its implications in the age of Industry 5.0.

4.2. Workstation Design for, with, and by Operators with Disabilities

It is understood that operators prioritise their involvement during workstation design and process allocation, since this would allow them to contribute their views and opinions and subsequently diminish the risk of problems encountered [2]. Therefore, the Operator 4.0 is to be involved in participatory design, professional development, and knowledge dissemination. Participatory design (PD) reaches its apex when knowledge (especially tacit or hidden knowledge) from a range of stakeholders is made communal, and thus stakeholders on all wavelengths may learn from one another [2]. This synergy is also shared by Oldfrey et al. [42], who highlight the opportunity of allowing people with disabilities to participate in the design of ATs, establishing a future where the end-user is also the designer or contributor. On a similar note, Mattsson et al. allude to the seven key principles of Universal Design to develop a work environment that is as inclusive as possible [7]. Thus, commonalities can be established in the remit of inclusive and participatory design. However, at present, based on the literature reviewed, there is an absence of such an approach. Industry 5.0 could therefore provide the proper platform to extend inclusive and innovative design towards smart, sustainable, and inclusive workstations as well [42]. Correia de Barros [43] rectifies by noting how inclusive design towards Industry 4.0 workstations calls for full participation of the ultimate end users during all stages of design. Rightfully so, Katiraee et al. [5] compel researchers to embrace a human-centred approach, including and deliberating with the end-users of such personalised workstations throughout the whole design cycle. Although not explicitly stated, such a recommendation simultaneously nods towards socially sustainable practices and therefore upholds Industry 5.0’s goals. It should also be noted (although not mentioned in [5]) that future workstation designs should also confer priority to sustainable decisions. These could manifest through selecting Industry 4.0 technologies that are versatile, modular, and that can thus adjust to multiple different users for tasks of levelling difficulty. The notion of considering task difficulty in task allocation is exemplified in the works of Kildal et al. [44]. For the same human–robot task, two diverse routes are presented, and it is up to the operator with cognitive disabilities to choose the preferred option (either one that grants more autonomy to the user versus a more guided, assistive channel) [44].

4.3. Balancing the Task-Skill Dynamic for Operators with Disabilities on the Shop Floor

On an equivalent wavelength, difficulty in allocating tasks to operators (with or without disabilities) may stem due to the operator’s age, gender, anthropocentric measures, and type or level of disability [5,24]. Secondary considerations encompass previous shop floor experience; exposure to noise, vibration, and ergonomic hazards; and cognitive aspects such as the operator’s motivation [45]. The latter determines productivity levels, health and safety, a potential escalation of work breaks and process times [45], and a re-evaluation of demand and potential costs. Product quality and the associated risk factors [5] are also two anchoring considerations in proper workstation design, particularly when a human-centric manufacturing approach is undertaken. Accordingly, Araújo et al. [46] employed a heuristic assembly line balancing problem (termed ALWABP) to strike a balance between employing operators with disability on parallel workstations, and the technical and financial requirements.

Litwin et al. [24] affirm that upon appropriate assignment of workers, no stark discrepancy will emerge in the times taken by operators with disabilities vis-à-vis those without. It all boils down to understanding the features, strengths, and weaknesses of each individual [5,24]. To exemplify, operators with intellectual disabilities were more productive when assigned to the nearest workstation to compensate for any sporadic breaks (without incurring supplementary 71travel times) [24]. This actuates another decision to be taken during work assignment, that is, whether to opt for fully manual labour, fully automated labour, or strike a balance between the two. Research has pointed out that opting for the latter broadens the possibilities for an operator with disability [47], provided that the user can switch between the two modes (manual and automatic) based on their temporal needs (for example, occasional fatigue due to medication). In addition, the total time taken for a task to be executed on a workstation is not entirely a product of the respective operator but also encapsulates the designated equipment and the task assigned itself. This may cause a rippling effect on other parameters such as the total cost.

In the age of Industry 5.0, which is anchored in human-tailored design as early as on the onset of task allocation, researchers have lent attention to in situ training in congruence to the task–operator skill dyad. A modular approach should be implemented to provide equal opportunities, grounded in one’s skills and task portfolio. In situ learning can be deployed in various applications and for diverse users, ranging from straightforward and safer interfaces for novel operators, culminating in more demanding modes for operators who would like to push themselves further [2,40]. A smart workstation would permit adaptable [9], guided, customisable, step-by-step tasks for each operator, irrespective of the abilities possessed [2,20,37]. This simultaneously nods towards a more socially sustainable manufacturing lens and thus conforms with the scope of Industry 5.0, where a “personalized co-learning and co-exploration experience” [48] (p. 614) is sought after, distinguishing discrepancies in “gender, anthropometrics, physiological and psychological” [48] (p. 614) human attributes.

Herzog and Harih’s work [49] can be deemed as a noteworthy outlier among the presented research bracket. The authors have attempted to promote the inclusion of individuals with physical disabilities by devising an assistance system grounded in ergonomic decision making. The proposed platform permits task-reassignment by matching the operator to a pre-available workstation that would cater best for the user’s physical limitations, strengths, and abilities (with emphasis on ergonomics). Despite such an innovative approach, a point of contention with this work is that the only “restrictions” captured by the tool are those related to lifting, movement and “other restrictions” such as hearing [49] (p. 228), with no mention of, say, intellectual or cognitive limitations. Another dispute is that despite the proposed system being simple and straightforward to use (thus can be used by any designer); choosing an individual from a database containing only “the most common disabilities that are relevant to a manufacturing environment” [49] (p. 227) could, in hindsight, be perceived as setting a hindrance to the operator’s ability to work. Nonetheless, Herzog and Harih’s approach [49] has laid a foundation for future research in this direction and should not be undermined.

4.4. Adapting, Accommodating, and Adjusting Workstations to Suit Operators with Disabilities

Industry 5.0 detaches from a fully techno-centric manufacturing shop floor towards a more holistic one, inevitably bringing forth potential adjustments needed to achieve this direction [5,9,10,12]. A diversified pool of workers may govern the technological equipment opted for when designing a new workstation, as opposed to simply following standardised procedures. This becomes necessary when encompassing workers with disabilities, warranting individual and custom workstations on a case-by-case basis and fostering the need for “smart and configurable workstations that are able to adapt to worker’s needs and capabilities” [5] (p. 3255). This is agreed upon by Mandischer et al. [19], who remark that despite a robotic work cell design being able to suit the operator’s needs, the robot itself may not be as apt in adjusting to stochasticity (such as dynamic and radical human needs [50]), incurring additional costs that deter manufacturing employers. In attempt to eliminate such challenges, the authors [19] used mathematical modelling to appoint tasks based on the operator’s contemporary skills profile and to amplify the work cell’s adaptability to the user through sensors that can observe modifications in the user’s behaviour. Similarly, each operator within Villani et al.’s study [51] is appointed an individual profile based on a pre-defined skills set that can be transposed towards adaptive interaction.

One critical observation from Katiraee et al.’s study [5] is the authors’ claim that it is challenging to gauge a worker’s level of impairment in production systems, considering that the majority of employees are “generally just classified into two groups: disabled or not disabled” [5] (p. 3256). Other scholarly work, however, fervently disapprove such remarks, pointing out that such segregation is not only unjust, but incorrect. Mandischer et al. [19] go as far as denoting how “able-bodied workers are PwD [people with disabilities] without limited capabilities” [19] (p. 1118), and that including operators with disabilities will reap opportunities for other individuals, such as the elderly, injured workers, or those workers who have acquired an impairment due to an illness. The authors [19] disregard the binary segregation introduced earlier (disabled vs. not disabled), but rather exhort an “ability” approach, whereby the operator is depicted as a collection of abilities and limitations. This perspective facilitates human–robot modelling since the operator’s limitations can very easily be delegated to the robot, improving inclusion on the shop floor.

Moore and Williams [52] and Lancioni et al. [53] assert that different levels of IDD severity warrant different needs: operators having a slight degree of IDD were entrusted with the operation of heavy machinery, whereas operators on the severe end of the IDD spectrum constantly necessitated the assistance of other employees. A noteworthy reflection was highlighted in view of operators with moderate IDDs, who carried out the work intended, provided that ample words of encouragement sufficed to keep the operators engaged in the task [52]. These observations were extracted from the context of a manufacturing facility and proved to be key for the AIDA social assistance co-robot framework proposed by the authors [52,54]. On the same wavelength, Lancioni et al. [53] integrated a remotely activated radio-linked praise system (three-word statements commonly used for encouragement) with optical sensors in order to prompt two men having multiple disabilities to manually assemble swivel wheels, whilst cutting the need for constant breaks. The authors reported affirmative feedback from this implementation, with both men successfully navigating through the assembly process, retaining accuracy and continuity. In a later research study, Lancioni et al. [55,56] re-opted for optic sensors to detect eight participants with multiple disabilities and issue verbal cues depending on the movement of the operator, such as “arrived” [55] (p. 150). Research marks the deployment of contingency measures and motivating comments to drive and retain the participants’ attention and steadiness, as well as improve their speed [11,56]. Importance was also devoted to physical mobility and rotation in both these works [53,55].

4.5. Key Takeaways from Section 4

This section elicits the profound lack in comprehending what designing workstations and shop floors for different abilities entails. There is obscurity on how a harmonious thread can be woven between adhering to the conventional requirements of the ever-evolving shop floor whilst upholding the needs of operators with disability. This balance would demand a thorough understanding of one’s physical and cognitive strengths and weaknesses and understanding how these interplay with the manufacturing tasks at hand. Consequently, future work (both scholarly and in industrial practice) should endeavour to resolve the workstation layout and design dilemma to fully accommodate the needs of people having multiple disabilities, as opposed to applying a binary segregation (either disabled or not) and potentially developing fruitful opportunities for all on the shop floor.

5. Industry 4.0, Industry 5.0, and the Impact on Inclusive Manufacturing Shop Floors

The literature has evoked the occupational barriers that could be confronted by individuals with multiple disabilities. Daily tasks that call for a multitude of steps are considered to be potentially challenging to certain individuals owing to limited reactive and recall abilities, alongside constrained motivation to manipulate objects and put them to use [57]. In order to provide aid for individuals with such potential limitations, specialised human supervision and staff assistance is often the answer [56], yet depending on the circumstances, such assistance may seem to be impractical if employed daily [55]. This may be the current snapshot within manufacturing companies employing individuals with disabilities. Consequently, alternative solutions are called for, especially within the age of Industry 5.0 where human-centred, socially sustainable manufacturing and operator empowerment are grounded alongside prominent Industry 4.0 technologies. Mark et al. [18] reiterate that Industry 4.0 should mutually complement the operator’s fortes rather than replacing or inflicting stress on the operator. There is an appreciation towards the window of opportunities that has been broadened through the implementation of Industry 4.0, pertaining also to individuals with disabilities who have previously encountered hurdles during employment. Equality and job satisfaction are the hallmarks of such a change, followed by the provision of employment even in the manufacturing domain of society, labelled as Production 4.0. Kildal et al. [58], for instance, note how Industry 4.0 aids people suffering from cognitive disabilities by assisting them in tasks that would otherwise prove too challenging.

5.1. Investigating Industry 4.0 Assistive Technologies for the Shop Floor

Although this study converging on operators with disabilities, the prospect of employing Industry 4.0 assistive technologies (ATs) for any operator (with/without a disability) should not be dismissed. Indeed, assistive technology may supplement additional relief and aid to operators, from the simplest task (such as understanding how theoretically distinct products from smaller lots are to be assembled) to the most arduous task. Researchers have also distinguished ATs based on their application, such as “intelligent assistance systems” [41], which are apt in surveying the user’s behavioural changes during the task and moreover their ability to adapt to these changes. Consequently, the application of IATs includes the creation of intelligence prompting systems [41] and can be of extreme value and assistance within the manufacturing shop floor context. As shown in [50], an array of computer vision and AI algorithms (referred to by the authors as cognitive assistive technologies or CATs) monitored the operator with IDs throughout the completion of an assembly task. The system would then feed on the user’s behaviour to decide which cues to present, and at which step to prompt them, bringing about an 85.7% completion rate without the need for any assistance. Hüsing et al. [59] recognise that ATs only live up to their name and scope as long as they tailor to the needs of their end-users, their abilities, and struggles. In this lane, the authors [59] propose a methodology that specifically fosters assistance towards people with physical disabilities (particularly those with musculoskeletal disabilities that hinder muscle coordination and motor skills). Doing so, the authors [59] solidified their claim that ATs for people with disabilities should boil down to each individual user’s needs. Mark et al. [18] categorise ATs into a triad comprising the sensorial, cognitive, and physical aspects. The most prominent ATs are accumulated in

Table 3. Summary of sensory, cognitive, and physical assistive systems, compiled from diverse publications within this review.

Assistive System Category	Technology	Purpose of Technology for Operators with Different Disabilities	Reference
Sensorial assistive systems	Warning lights	Exemplar for people with hearing disabilities, since such warning lights can allow for timely reaction.	[18,50,60]
	Audible signals	Idem to above purpose, with the exception that audible signals are suitable for people with vision disabilities.	[18,50,53,60,61]
	Visual instructions	Ideal for people with vision disabilities; useful to convey warnings easily and reliably.	[11,18,50,60]
	Smart watches	When amalgamating sensors such as heart monitoring sensors, real-time evaluation of the operator’s wellbeing is just a tap away.	[18]
Physical assistive systems	Exoskeletons	Physically worn by the user; practical for people with physical disabilities; promotes mobility to upper limbs, forearms, and whole-body movement.	[18]
	Collaborative robots and attachments	Integrated within the workstation to act as an extension of the operator, to sustain the operator during heavy load lifting—stimulates an ergonomically-conscious environment. Also suitable for tasks demanding a high degree of precision.	[18,59,62,63]
	Virtual reality and head-mounted displays (HMDs)	Provides hands-free control for people with restricted dexterity and motor skills.	[35,62]
Cognitive assistive systems	Projectors and thermography technology	Used to avail operators of guidance during assembly tasks, whilst ensuring that an acceptable quality is simultaneously attained.	[11,33,36,44,58,61,63,64]
	Digital supportive services such as contour visualisation	Feasible for people with cognitive disabilities and provision of workload relief as opposed to having to recall everything by memory. Clear outlining of contours and instructions was also recognised as a beneficial implementation.	[11,33,36,63,64]
	Collaborative robots and attachments	Supplement guidance during cognitively demanding tasks, such as laser-pointing and tracking.	[44,58,65]

, in concurrence with literature that alludes to each respective category. Each category attempts to utilise the advantages of Industry 4.0 to aid operators with disabilities.

The current literature is neglectful of frameworks that establish a steady correspondence between different ATs, heterogenous operators, and the involved work. Fittingly, Mark et al. [66] propounded twenty-three skills that diverse groups of operators (elderly, unskilled, workers with disabilities, workers with health-compromising risks) may necessitate when working on the shop floor, such as the ability of sight, ability to focus on the task at hand, and the ability to solve problems with ease. These skills are then administered to the AT that best upholds that skill. Thus, this addition maps the desired direction forward. Having said that, there is obscurity in the rationale behind the interdependence between ATs and the required skills. For instance, there is no implication of how and why certain ATs were allocated to particular skills, as well as the justification behind the allotted numerical ranking between the pair.

There is also a notion of segregation present, whereby each AT is correlated to a skills parameter on its own. A simple illustration of this observation can be made for the AT labelled “eye tracking” [66], which is ascribed a score for six of the twenty-three skills. One might pose the question of whether the attributed relationships would have changed had the AT in context been employed in conjunction with other ATs on a workstation, and hence, one AT would influence the next, leading to potential ramifications on operators with different disabilities. Indeed, Leng et al. claim that within the Industry 5.0 paradigm, no single technology should be considered independent from the next, but a “reorganized symbiosis” [29] (p. 291) should be ascertained. This was also confirmed through the study of [32]. Furthermore, Lancioni et al. [53] claim that the choice of technology should stem from the degree of support supplemented by said technology, the expenses to acquire and maintain the AT, but also the level of usability of the AT depending on the individual. Priority must also be attributed to selecting the most appropriate interface modality [67] and the way that information (both input and output) should be presented to the operator, so that no overlap arises between one AT and the next. Control is another vital element to enmesh during the onset of workstation design [10], elevating the relevance of human-centred design.

Moreover, researchers have also devoted studies to understand how diverse users can exploit their available senses to issue commands when working with a diverse portfolio of ATs. Indeed, Lu et al. [48] gauged the importance of selecting the most appropriate sensorial communication mode (spoken, gaze, gestures) depending on the level of human–machine comprehension preferred. Such potentials encompass (but are not restricted to): tilting of the head and neck [68]; gaze and eye-tracking [69]; speech [62]; and gestures [2] to retain autonomy of the system by a person with disabilities. Lu et al. [48] deem gestures and action detection as the most applicable for human goal and action understanding by the system. Gesture detection and interpretation by the system permits observation of the operators as they carry out a given task, and it subsequently aids operators of distinct abilities to successfully execute said task [2]. Contrarily, Kildal et al. [44] underscore that the participants recruited to assess their HRI task (individuals with cognitive disabilities) widely opted for speech inputs and outputs as opposed to gesture-based modalities. In fact, the authors observed that gestures were never the go-to option for their sample size but were rather coupled up with speech—the main modality of choice. This highlights a research vacuum within the umbrella topic of multi-modality. There is no clear-cut consensus between academics in comprehending the preferences of people with different abilities and the most suitable technology for the specific modality, task, and individual.

To support the above, it can be observed that each of the four mentioned modalities constitute their own bundle of challenges. For gaze, these concern the intensity, duration, and fixation of the gaze and distractions such as other workers in the visual peripheral [67]; for head and neck, the user may not be able to pitch, roll, and yaw their head properly due to their disability; and for speech, the AT may not recognise voice commands immediately, especially if the user’s voice is frail or incomprehensible due to the disability (in this case the disability was multiple sclerosis) [62]. All these hurdles raise the need for modality alternatives that adopt and transpose the techno-centric approach of Industry 4.0 to a human-cognisant Industry 5.0 application. Smart manufacturing technologies such as virtual and augmented reality for inclusion are explored accordingly in Section 5.2.

5.2. Augmenting the Workspace of Operators with Disabilities on the Shop Floor

Sustaining a vast choice of modality options for people with different disabilities can also be attained through proper application of two crucial Industry 4.0 driving technologies—virtual (VR) and augmented reality (AR). Budziszewski et al. [70] constructed multiple arguments in favour of virtual reality’s applicability both in manufacturing as well as to cater for a heterogeneity of workers. Primarily, VR embraces a safer learning space at a reduced cost [70,71], proving to be exceptionally vital to train operators (especially those with a disability) who may have little to no experience working with machinery or on the shop floor [2]. Although this observation is true in its essence, there is ambiguity in relation to how VR would affect individuals with comorbidity, namely, the cognitive effects that exposure to VR for a prolonged period would incur on the user. This warrants further research in pursuit of understanding which Industry 4.0 technologies are optimal for which users’ abilities and needs. To elaborate further, Budziszewski et al. [70] were also able to assess the user’s interaction and operation on the workstation, in tandem with the operator simulating the work. While this is a significant benefit of VR for an inclusive workstation design, there may be an underlying assumption that the worker has already acquired the knowledge of how the task should be carried out, which is neither always the case nor always possible.

A further superiority of VR for people with disabilities on the shop floor transpires in the fact that no tangible workstations need to be physically deployed or modified, but rather they can be manipulated with ease using the VR software in context, minimising time, costs, and resources [70]. Despite not being addressed in the works of [70], it can be underscored that such a plus-side of VR is congruently upholding environmental and economic sustainability. Furthermore, the former observation (safety in VR and a better training experience) highlights how if employed with purpose, VR also conforms to the social dome of sustainability. Indirectly, this is in line with the principles of Industry 5.0, bridging the sustainability triad with human-centricity in a resilient manner. Research should thus dedicate effort towards such an exercise—mapping how and which Industry 4.0 technologies can be employed to uphold the novel Industry 5.0.

Simulating real-life scenarios a priori to implementing drastic changes on the shop floor is the optimal way to appraise diverse possibilities without endangering anyone on the shop floor. Despite such benefits, simulation and virtual spaces may raise concerns on the difference between the user’s natural interaction with the workstation and the VR experience (where the sense of touch is not possible to be executed, causing unfamiliarity with the tangible experience) [70]. One consequence of this is that employing VR during workstation design may prioritise the workstation itself (once again conforming to Industry 4.0’s preset techno-centricity). Such tunnelled focus may overlook human-oriented aspects such as touch sensitivity or the level of fine or gross motor skills needed when carrying out tasks involving very small components [70]. Litwin et al. [24] reverted to Vensim [72] and FlexSim [73] to present a discrete event and system dynamics simulation which investigates how operators with disabilities on the shop floor would map on the assembly line and the company. The study involved both operators without disabilities, as well as those with mild intellectual impairment, so that a paradox could be drawn between the two. Simulation involving operators with disability may lead to other underlying hurdles since no virtual simulation may predict the sporadicity of such operators (for example, the unplanned breaks addressed by Litwin et al. [24]).

Comparison is the key to wiser, and faster, assembly time. This is reiterated in multiple literary works, in pursuit of identifying which type of instructor assistance fares the best within the provided manufacturing context. This is especially true when investigating situations which involve workers with a disability. Kosch et al. [60] investigated the preferences of operators with cognitive disabilities when it comes to dissemination of error-warnings on the shop floor. A comparison is struck between visual (employing red light projection), auditory (attributing sounds to an error), and haptic (transmitting pulsations through Wi-Fi-enabling hand gloves) modes of raising error awareness, resulting in the visual projection ranking highest and the auditory depicted as disturbing. Another observation ensued from this work is that none of the participants (individuals with cognitive disabilities) responded to the errors in the same manner, affirming the notion that one size does not fit all. Notwithstanding, colour-coded visual projection may account for unforeknown obstructions for operators who are colour-blind, as was conveyed by Jost et al. [74] and Villani et al. [40].

Funk et al. [36] reflect upon task allocation and propose in situ projected instructions through industrial augmented reality (IAR) to permit for more dynamic tasks carried out by operators with cognitive disabilities. It must be underlined, however, that the tasks delegated comprise assembling LEGO bricks (such micro tasks do not emulate the manufacturing shop floor environment, where more composite tasks are designated). In situ projected instructions relieve operators who have a shorter working memory of having to switch between reading instruction manuals and then having to transpose those instructions to the task at hand [11,75]. Conversely, the instructions are projected in such a way that they are presented in congruence with the task. In this manner, Funk et al. [36] enlist augmented reality as one of the enablers to inclusivity within manufacturing tasks. Their setup reaped a reduction in assembly errors by half, and assembly time was reduced to a third. Likewise, Aksu et al. augmented the working environment for people with minor intellectual impairments through a mobile application, guiding the operators through the sequential task of replacing broken drill heads [75].

AR yields a range of potential applications within manufacturing, such as assistance towards maintenance [71] and logistics (leaning towards the techno-centricity and production-geared purposes), together with broadening the “pedagogical” [35] (p. 436) view pertaining to guidance during vocational training [11,33,35,71]. Büttner et al. [35] exhibited further potentials of AR in promoting inclusivity not solely for operators with disabilities but also for unskilled and inexperienced workers. Segura et al. [71] draw attention to how AR grants the operators the ability to view and follow information in hands-free modality and alleviates operators from the cognitive burden of having to constantly refer to separate instruction manuals. Such benefits could be favourable considering an operator who gets constantly distracted [51], or who is unable to carry heavy screens due to restricted dexterity.

In accord, Jost et al. [74,76] opt for AR projection as the preferred technology in pursuit of a contour-based instructive counting aid, PARTAS, for individuals who find difficulties in working with and recalling numerical connotations. This study should be commended for reflecting upon an often-overlooked aspect within manufacturing—working with numbers, measurements, dimensions, and quantities [74]. The ability to count and understand the meanings represented by numbers is often taken for granted, yet this would be an insurmountable hurdle for operators with dyscalculia (incomprehension of numerical meaning). Funk et al. [39] justify taking on pictorial full-size contour-based instructions owing to their ease of “transferring information” [39] (p. 1) to operators with disabilities, which is a crucial consideration when a new skill or task is being imparted. Comparably, Aksu et al.’s provision of video-based assembly instructions rendered an independent and positive learning experience for operators with mild intellectual disabilities on the shop floor [75]. Proper knowledge dissemination is a rudimentary prerequisite in all capacities of learning, thus rightfully warranting equal attention upon the manufacturing shop floor, especially considering the heterogeneity of workers present (such as the fact that not everyone learns in the same manner or works at the same pace [46]). Although often overlooked, workstation design in the remit of a socially sustainable Industry 5.0 should therefore factor in the aforementioned concern [29].

Despite the advantages denoted by [36,76], similar case studies within proper industrial contexts have not been significantly implemented (due to the general preference for implementation of such studies within sheltered workshops such as in [76]). This gears the researcher to conduct in-depth practice on the manufacturing shop floor for improved outcome of results and longevity, as is the case in Aksu et al.’s study [75]. Such a suggestion spans from an observation entrenched in some of the case studies centring around sheltered workshops. Here, it can be deduced that human adaptability is amplified whilst productivity is placed on the back burner. This is echoed in scholarly excerpts such as, “In this context, the system does not attempt to increase productivity” [76] (p. 217) and “PARTAS features a reduced level of required positioning accuracy of parts” [76] (p. 217). Whilst human centricity is indeed one of the three hallmarks of Industry 5.0, it is essential to strike a balance between productivity, flexibility, practicability, and quality [74], whilst concurrently respecting the operator. Only then will a seamless transaction ensue between Industry 4.0 and Industry 5.0, without pedestalling one at the expense of the other.

Simões et al. [61] outline the pivotal effect that technology has rendered on empowering operators with disability, notably throughout the training stage. The authors point out the scarcity of literature that puts the skills necessitated during training into perspective and thus proceed to develop a fully interactional system that employs sensors to uphold the mental well-being of operators with cognitive impairments. With the introduction of cross-reality (XR), the authors witnessed a rise in productivity, as well as acceptance from the operators. Analogous positive outcomes are reiterated in other scholarly works, such as that of Lancioni et al. The authors of [56] report on how individuals with multiple disabilities felt independent and coherent with the assistance of technology during a water pipe assembly task. On a similar wavelength, Villani et al.’s work [4] caters for an inclusive training methodology that onboards virtual reality as the main assistive tool for user empowerment and timely learning. Having said that, this methodology is limited to human–machine interfaces and not the whole workstation design (including any external influences). Conversely, Budziszewski et al. [70] coupled up ergonomic human simulation and modelling with virtual reality to depict a fully immersive experience of the whole workstation, annotating how an individual with limited mobility would interact with such a workstation.

Simões et al. [61] acknowledge how diverse people have their unique limitations (visual, audio, and manoeuvrability), posing different skillsets (for example, it cannot be assumed that everyone is literate), which reflect on manufacturing parameters such as the time taken to complete a task. The authors also echo how multiple disciplines (such as psychology and cognitive sciences) must be intertwined with that of engineering when attempting to understand how diverse people learn, work, and train, challenging the status quo on the shop floor. This research work also alludes to routine practices such as consideration of the surroundings (light, temperature, audio levels, ergonomics, and safety), but also factors in personal elements, namely, current health condition and the user’s skillset. Despite these crucial factors, the authors overlook additional criteria such as the expenses incurred to set up the proposed training station (tying in with the firm’s economic sustainability) as well as whether the station is at the end of the line and thus would require additional inspection and checking to ensure quality.

Disability vs. Simulation

Primarily, it must be recalled that operators with disabilities may encounter hurdles in rudimentary activities, and these will pervade their performance on the shop floor; for example, an operator with upper limb limitation may work or move slower than others. Simulating such a scenario raises concerns on what rate (in terms of time) should be assumed to such an individual. Budziszewski et al. [70] employed two participants with limited upper movement to mirror how individuals with physical impairments may be simulated. This was attained by utilising the standard human model followed by inactivating the permitted activity in the upper extremities (restriction of movements). On the other hand, one may revert to a second example: a person with an intellectual disability may have a diverse reaction time to an accident than a mainstream operator, as well as a higher risk of unpredictability and pauses which simulation cannot predict in advance. Ultimately, simulation may need raw data collection to base a concrete idea on, and this is not possible if there are little to no operators with different disabilities actually carrying out tasks on the shop floor. Indeed, human simulation modelling was employed throughout a virtual reality exercise by Budziszewski et al. [70] during the design of workstations for operators with mobility impairments (upper limb impairments), as opposed to intellectual impairments.

5.3. The Uprise of Robots Collaborating with Operators with Disabilities

As elaborated in Section 5.2, simulating scenarios for inclusivity comprises, but is not restricted to, AR and VR, also factoring the deployment of collaborative robots (cobot/s) working along operators with disability. Human-centred collaborative design is pivotal for a number of studies [40,54] investigating how collaborative robots can be adapted towards assisting workers with disabilities. The AIDA social robot prototype [54] has tapped into multiple aspects concerning manufacturing, such as permitting operators with IDD on the shop floor to request (through the robot’s interface) amendments to their workspace, namely, to eliminate obstructions in their way and enhance the overall aesthetics of the space. AIDA is also forecasted to engage with shop floor supervisors when it deems that the operator is fit enough to learn a new task (prompting job rotation for onboarding of a myriad of novel skills). Other studies have also presented frameworks (such as the INCLUSIVE framework by Villani et al. [40]) which uphold assisting operators with disability on the shop floor through system adaptation based on the operators’ abilities and present needs, timely training of operators, and also keeping account of the human’s abilities and any changes in between. Weidemann et al. [77] demonstrated the significance of determining the level of human difficulty before assigning a cobot in conjunction with operators having a multitude of disabilities. According to the authors, only upon comprehension of this level of individual challenge can the most suitable assistance be leveraged by the cobot. Consequently, the authors devised a software tool that is able of mapping this extent of difficulty encountered during the specific task, based on the IMBA (Integration of People with Disabilities into the Working Life) and O*NET (Occupational Information Network) workflow examination tools.

Multimodality, user-centred design, and adaptability with respect to operators with disabilities on the shop floor were addressed by many scholars as elaborated in this review. Having said that, however, there is a propensity for multimodality to be confined solely to the presentation of instructions (either visual, projection-based, verbal, audio, or contour-based instructions), without fully articulating how multimodal options simultaneously influence the application and potential of different technologies. Stöhr et al. [63] underscore this by presenting how a collaborative robot’s versatility and multimodal potential can be exploited to assist operators with visual and cognitive disabilities. This research work [63] depicts the proper transition that should ensue between Industry 4.0 and Industry 5.0, owing to elaborate detail provided even in terms of system architecture and how the assistive robot would integrate with other systems, networks, and flows on the shop floor. Such detail was rarely encountered in the other research work perused for this review.

Drolshagen et al. [65] scrutinised the influence of a KUKA LBR iiwa 7 R800 collaborative robot (cobot) arm in the pursuit of inclusivity, commencing by acknowledging that the current literature remains neglectful of how people with disabilities perceive working alongside a cobot arm. It is thus presented how collaborative robots can assist people with physical and motor limitations in, for example, lifting heavy objects, facilitating reach, and autonomously carrying out specific tasks [12,18,29]. Additional assistance is also provided to people with disabilities in their daily endeavours such as in imparting rehabilitative movement and providing ample training [65]. This becomes focal in view of the impediments confronted due to smaller lot size production in manufacturing companies, where variety in each individual production lot equates to changes in assembly and updates in the training provided.

It is worth noting how different disabilities shape versatile human–robot collaborative (HRC) dyads. For instance, if the operator has a physical disability only (and thus has full control of their cognitive abilities), then a cobot is the perfect partner to delegate physically demanding tasks to [68]. On the contrary, should the operator have cognitive disabilities, the roles are now reversed. The ambidexterity of the operator is put to their advantage, whereas the cobot is entrusted with the cognitive load of, say, identifying for the operator which cable to connect. A cobot is rightly labelled as a social “actor” [47] as with appropriate planning; it can pave the way for diverse users in an environment that may ostracise people with disabilities. The work of Kildal et al. [44,58] upholds the latter by installing laser pointers to the end-of-arm-tooling (EOAT) of a YuMi cobot to guide the operator, as shown in Figure 5. Such a setup excludes unnecessary instructions that might confuse operators yet is still sufficient to grant empowerment to the operator.

Although not addressed in [58], a third possible case scenario entails having operators with both physical and cognitive disabilities on the shop floor. Consequently, this would compel a completely new outlook towards a revised implementation of cobots that would reap the highest advantages from both parties. Another study pertaining to employing collaborative robots as part of the AT family is that of Hüsing et al. [59], where the authors devised a novel methodology for inclusive manufacturing. The authors suggest organising each task into smaller fractions, followed by a thorough evaluation of the requirements for each fraction (“capability-based workplace requirements”) and then concluding by drawing parallels between the individual worker’s skills (“individual capability profile”) and those of the cobot. Upon further scrutiny, it can be acknowledged that all the above studies elicit one fundamental aspiration, that is, appointing cobots for people with disabilities. Light is shed on how the versatility demonstrated by cobots can be exploited to mimic a supplementary operator and thus occupy some or all labour-intensive tasks. Notwithstanding, introducing a cobot might unintentionally provoke cognitive stress for an operator with, say, a cognitive disability, which in turn may lead to unanticipated accidents or events. Consequently, Section 5.4 elaborates on how key enabling technologies (such as cobots) can work concurrently to uphold both physical and cognitive well-being in a smarter, more sustainable, inclusive manufacturing environment [78].

5.4. Cognitive Ergonomics for Inclusivity in Industry 4.0

Researchers are acknowledging the underlying cognitive transitions that may need to take place between human and machine, specifically, the level of human acceptance and the corresponding interactions [7,12,29,37,40]. This complements the root principles of Industry 5.0, that is, to adopt human-centred manufacturing that surpasses physical human factors such as safety, but upholds “a higher humanistic level, such as cognitive and psychological wellbeing” [48] (p. 615). Lu et al. [48] appoint crucial variables to consider on the onset of collaborative workstation design for the mainstream operator: (a) the level of two-way trust (human and machine/s), (b) ease of usability, (c) potential stress and frustration due to workloads, (d) envisaged enjoyment and corresponding levels of satisfaction, and ultimately (e) the human’s acceptance of the workstation. This range of elements is also validated through the study by Villani et al. who based their system’s adaptability on, for instance, the present level of strain undergone by the operator [51]. These factors may manifest themselves differently in diverse workers, based on the individual’s abilities and limitations (physical or cognitive). This pinpoints to a twofold struggle that needs to retain equilibrium: prioritising the operator and HMI experience by mapping variables (a)–(e) to the personal needs, yet simultaneously designing a workstation that can satisfy the daily production targets. The former is yet obscure and generalised and, hence, warrants investigation on how operators with different disabilities respond to variables (a)–(e): how concerns are channelled and how user-centred design can meet everyone’s needs. Ultimately, it is vital that both human and machine achieve a high level of mutual trust [68].

The level of autonomy and control over the robot system should acquire priority, especially for operators with disabilities, who may encounter physical difficulties in accessing controls but also cognitive hurdles in issuing the right commands. Arévalo et al. [68] converged their research on people with motor disabilities; their potential working with collaborative robots; and equally, the obstacles confronted in pursuit of this goal. Notwithstanding, administering a cobot in a sheltered workshop acts as a scaled-down substitute to running the experiment on the actual manufacturing shop floor, so the dynamic shift in the two environments prevents a true reflection of how cobots would fare with operators with disability on the shop floor. For this reason, Litwin et al.’s [24] study should be praised for using a proper manufacturing environment as the basis of their simulations of operators with disability. Such a route was also adopted by Villani et al. who validated their INCLUSIVE framework [40] through practical case scenarios that capture the reality of shop floors, their operators, and the actual tasks to be delegated, as opposed to implementation within sheltered workshops.

One may, however, reflect on how actual trust differs from human to human, and how such a consideration influences the choice of technological components for inclusive system design. This matter is furthered when technologies such as cobots where mimicry can go both ways (human to robot vs. robot to human). Such engagement takes HRI to the next level, yet extra caution should be applied when the system is intended for people with disabilities. This recommendation stems from the extent of emotional leverage that engaging with a robot may impose on the operator with disability. This is of paramount importance due to differences in perception, situation awareness, processing of events, predictability, interpretation of current status, and comprehension that would have otherwise been overlooked depending on whether the worker occupies the operator or the supervisory role [37,68]. A spotlight should be shed on the ever-growing notion of anthropomorphism in manufacturing, whereby machinery (such as collaborative robots) is dressed in attributes that mimic those of humans. A case scenario would be the ROBOTIQ 3-Finger Gripper presented in Figure 6, whereby resemblances can be clearly drawn between a human hand and this gripper. This aspect of anthropomorphic functionality (for example, a gripper impersonating a human hand during a pick-and-place) may unknowingly evoke social and cognitive triggers, such as animacy (in the sense that the robot will socially appear as “animated” instead of merely carrying out a designated task).

Despite the exhibited potential of HRI for operators with disabilities, the actual interaction and control allocation warrants significant thought, owing to fact that the operator’s inputs will be mapped to a “machine” that is respondent in the real world and whose actions may inflict on others. Kildal et al. [58] note that one hurdle to be leapt is designing cobot cells that can warrant the trust of the user, as well being able to facilitate information transfer. It is thus crucial to comprehend the end-user’s needs especially in HRC and thus re-evaluate the allocation of the tasks undertaken by cobots and humans individually [58].

Consequently, one might question whether Industry 4.0 technology is deemed as an intruder in one’s personal space as opposed to acting as an assistive extension of the individual. This matter is also subjective depending on the type and severity of disability in context, with the latter not being made clear by some authors in their work. Drolshagen et al. [65] conducted a study in a sheltered workshop to comprehend the effects of varying distances between the cobot and person with a disability in order to understand what influences the familiarity and trust between the two parties. It is imperative to bear in mind that sensory challenges may be present in people with disabilities, and thus an assistive robot could trigger sensory overload, revoking its applicability to sheltered workshops. Conversely, operators with other disabilities (for example, restricted movement) may not mind working along an assistive robot due to their familiarity and continuous use of ATs to ease their daily mobility struggles [47]. This embodies another challenge for HRI workstations, since trust is a subjective product of (but not limited to) knowledge and personal experience; thus, a heterogenous workforce warrants individual attention prior to design.

All these emotions have a direct influence on the performance and, most critically, safety of the human working with the cobot and vice versa [37]. For instance, under-trusting the cobot could result in sporadic human behaviour, whereas over-trust could result in nonchalant behaviour that endangers the user and others. Similarly, access to control for, say, a person with motor disability is much different than access to control for the mainstream operator. Assignment of control equates to the appointment of responsibility, and whilst advocating for operator autonomy [44] and control is the direction towards inclusivity, responsibility must also be taken when things go amiss and rapid intervention is required. Arévalo Arboleda et al. [47] stress this notion, with participants having motor disability in their study suggesting that the machine should be equipped with “sensors to detect automatically if something is not being done correctly” [47] (p. 4). The machine was also expected to perform appropriately in case of abnormalities, so in this study, preference of assigning full responsibility to the machine (the robot) was observed. However, fulfilling these requisitions is challenging, for which other participants in the study [47] urged the use of multiple accessible options for abruptly halting the system.

5.5. Key Takeaways from Section 5

Even though multi-modality provides a level playing field for any operator (with or without impairments, experienced or unskilled, old or young), potential pitfalls may be encountered. It is worth attuning to the consequences of these limitations, namely, operators with disability not being fully conscious of their concurrent mode of operation, resulting in distorted responses and confusion as to when a particular mode should be opted for against another; in hindsight, such misperceptions may incur additional unplanned accidents, or prolonged production time. Drawbacks related to time may also be caused when simplifying human–machine interfaces to cater for specific cognitive abilities, yielding an inadvertent increase in task time. This may thus lead to conducting trade-offs (choosing production parameters (such as time) over socially sustainable parameters (namely, operator well-being) or vice versa). Another observation adopted from the work of [61] is that one should not refrain from incorporating multiple modalities simultaneously (such as visual AND audio at the same time), owing to the reinforcing element that is hence introduced. The user can confirm that they have understood correctly due to multiple modes that assist in information input and output. In doing so, the cognitive uncertainties of not having timely feedback [20] is lessened. Korn et al. [64] acknowledge this concern by stating that at times even assistive systems themselves fail to provide the ideal human–machine interface (HMI) between the system and the persons with disabilities. Hence, it is not sufficient to stop at the requirements and parameters. Therefore, another criticism to point out is that despite addressing both requirements and parameters, the authors [80] have also not considered how the design parameters themselves would need to be modified and tailored to suit people of varied needs, instead of opting for a one-size-fits-all approach. Consequently, there is still work to be done on finding the optimal practical collation of devices that can suit a plethora of disabilities. A consecutive question would then be with regards to the choice of the devices and their conformity to the three sustainable pillars (social, economic, and environmental). This leaves the reader (for example, a potential employer) to ponder on which is the best route forward, as well as how inclusion can be sustainably attained through Industry 4.0.

6. Sustainability on the Manufacturing Shop Floor and Industry 5.0

6.1. Industry 4.0 for Sustainability, or Sustainability for Industry 4.0?

With today’s incessant resource consumption and depletion, coupled up with a colossal increase in CO₂ emissions and pollutants per annum, it is undoubtedly important that manufacturers foster a production perspective that is flexible and efficient in order to ensure longevity of resources. The amalgamation of industrialisation (and thus, Industry 4.0 and all its technologies) with sustainability has therefore given rise to a conflicting issue, which is the beneficial but potentially adverse influence of Industry 4.0 on sustainability. This calls for a mediation to be struck between the two so that the advantages of Industry 4.0 on all dimensions of sustainability should be also encouraged and thoroughly planned. Ghobakhloo [31] endeavours to depict Industry 4.0 in its transparency, elaborating on sixteen diverse areas that bind Industry 4.0 and sustainability together.

Table 4. Identification of sixteen areas that bridge Industry 4.0 and sustainability together.

Areas of Commonality between Industry 4.0 and Sustainability, as Displayed by Ghobakhloo [31]	Economical	Environmental	Social
Upscaling of business framework	✓
Cutting down on CO2 emissions		✓
Enhancing company profitability	✓
Adopting economic growth	✓
Resource management and energy efficiency	✓	✓
Consciousness towards the environment	✓	✓
Strengthening human resources	✓		✓
Prioritising efficiency in production	✓	✓
Provision of novel occupations	✓		✓
Decreasing costs to manufacture	✓
Augmentation of manufacturing flexibility	✓	✓
Adaptable production	✓	✓
Ease of production versatility	✓	✓
Upgrading safety control through cobots			✓
Managing supply chains	✓	✓
Enriching social consciousness			✓
TOTAL	13	8	4

collates these areas and proceeds to extract the pillars that most correlate to each of the sixteen categories.

Through this analysis, it is clear that Industry 4.0 is geared towards satisfying economic sustainability, amplifying on efficiency and productivity, with the result of extending the sustainability frontiers towards the economic–environmental stratum [12,31,78]. Only four out of sixteen of Ghobakhloo [31]’s identified topics partially concern the social sphere of sustainability, with solely two domains denoting how Industry 4.0 can amplify the quality of manufacturing jobs, secure operators’ safety, and heighten the current social welfare. The author elaborates on how Industry 4.0 technologies, “smart sensors, smart safety wearables” [31], can be exploited to exercise safety. For clarity, this may be exemplified amidst the context of HRC workstations, where the propensity lies for a safe HRC workstation to be vaguely labelled as “sustainable”, when it would only be addressing the social third. This is unfortunately avid in numerous research works, such as in [37], where despite its depth in terms of human-centred and safe HRC (social sustainability), the authors failed to touch upon the environmental consequences HRC may yield. Mention of the environment only alludes to the manufacturing environment itself (mainly through aspects such as ergonomics), as opposed to adopting a systems, products, and processes approach. Over and above, oftentimes, when correlating sustainability to Industry 4.0 counterparts, the relationship ceases at the detection of abrupt anomalies, and other opportunities for Safety 4.0 are overlooked, especially when attempting to comprehend how the same technology can be reverted for people with disabilities.

It is only in the very recent years that Industry 4.0 technologies and key enablers of Industry 5.0 are viewed from a multi-disciplinary lens [29]. A noteworthy observation raised is the perception of customers who consume “ultra-personalised products (UPPs)” [31] and customised products [10], produced with the engagement of Industry 4.0 technologies. Researchers and industry alike must challenge the established notion of “customers” so that the same approach and consideration is also instilled for operators with disabilities. Operators with disabilities have the same right to UPPs, even in the context of workstations tailored towards inclusivity and relief from both physical and cognitive workload. This aspect is fruit for thought and is harnessed as a dominant factor in the next section.

6.2. A New Outlook towards Sustainability in the Age of Industry 5.0

The new Industry 5.0 paradigm focuses on sustainable yet simultaneously smart resilient manufacturing methods and attempts to break away from what formerly revolved around “system-centric” manufacturing to prioritise “human-centric” manufacturing [9,10,48,78]. The latter employs Industry 4.0 technologies to attend to the user’s psychosocial attributes and needs. The authors of [48] conceived the Industrial Human Needs pyramid in which five tiers (safety, health, belonging, self-esteem, and self-actualisation) embody critical and fundamental necessities for human-oriented manufacturing, as depicted in Figure 7.

Such necessities should make no distinction between operators of different abilities and should be attainable for everyone, especially the top three levels (3, 4, and 5) which are more subjective and person based. As an illustration of this, one may note that aspects such as trustworthy human–machine collaboration can be enlisted as a contributor for Levels 2 and 3 [48]. Such a goal may be permissible through the creation of a human-oriented digital twin [48] that enhances the worker’s experience through a real-time, in-depth human–machine symbiosis, monitoring a diverse array of parameters ranging from physical to emotional collection of data, allowing personalised training for each worker and improving task allocation and factory layout.

The key issue within Industry 5.0 is comprehension of how vital Industry 4.0 components (such as a sensors, digital twins, big data storage, and artificial intelligence [9,10,29]) that were previously associated with monitoring production on the shop floor can be transposed to assist the operator instead to uphold the smart and sustainable belief. Consequently, it is worth pointing out the suitability of sensors in the quotidian activities of individuals with disabilities in attempt to create “self-operated prompting systems” and “automatic prompting systems” [41]. In this regard, the applicability of sensors should permit more than merely sensing the environment and observing the user’s behaviour but should also envelope safety considerations (in order to eradicate the possibility of false alarms, false positives, and false negatives; for example, issuing a prompt for Step 2 when the user is still in Step 1).

Mandischer et al. [19] employed exteroceptive sensors upon a sense–plan–act approach, wherein a robot observes the behaviour of operators with disability, permitting a timely reaction, response, adaptability, and task allocation. This exemplifies the full-circle moment that Industry 5.0 aims to obtain, with the present logjam being the scarcity of scenarios where Industry 4.0 technologies are employed for all three domains of sustainability concurrently. This observation can be further appreciated through Oldfrey et al.’s [42] research work, which is one of the few studies that has managed to coalesce three (disability + assistive technologies + sustainability) of the four disciplines that are rooted in this literature review. Nevertheless, priority is once again devoted to the environmental “climate change” [42] (p. 3) the and economic aspect of sustainability; however, alluding to “equitable reliable access” [42] (p. 1) indirectly kneads in the social domain of sustainability too. This outlines one of the research vacuums identified through this review, where at times, the reader is left to decipher and interpret where and how social sustainability would be presented, gauged, and factored among the more transparent environmental and economic aspect. Oldfrey et al.’s work [42] stems from attributing the circular economy approach in the context of assistive technology design and conceptualisation, as exhibited in the circular AT model of Figure 8.

In this domain, the circular economy (reducing, reusing, regenerating, repairing, recycling, and keeping ATs within the loop as opposed to discarding [42]) is viewed from a sustainable and resilient design point of view. Although not mentioned, such design principles could be taken forward and applied throughout the design of smarter and sustainable inclusive manufacturing workstations. Oldfrey et al. [42] have thus elicited that indirect links between disability (through life cycle of ATs) and environmentally sustainable manufacturing are indeed being drawn. Such momentum should, however, extend also in the direction of engaging a fully socially responsible and inclusive shop floor (not just ATs and AT production), the aspects of which are tackled in Section 6.3.

6.3. Socially Sustainable Considerations for Operators with Disability on the Shop Floor

Theoretically acknowledging the social sphere of sustainability within the manufacturing and production environment is not a novel concept (such as Romero et al.’s [25] pivotal study in 2015), where human-centred design, job satisfaction [40], and a flexible human–machine dyad are enlisted as prime contributors. Nonetheless, proper practical implementation and guidance of social sustainability in manufacturing has not caught enough momentum yet [32]. Authors such as Villani et al. [4], Leng et al. [29], and Majernik et al. [81] present the importance of striking a balance between social sustainability and traditional manufacturing parameters: “optimal flexibility, agility and competitiveness of highly customised production on the one side, and, on the other side, a sustainable effort for all workers” [4] (p. 1). Emphasis is to be exercised on the word “all”, showcasing potential for more symbiotic work whilst steering away from what is considered the norm on the shop floor. Industry 5.0 researchers are also advocating for further practical considerations of the relevant sustainable development goals (SDGs) within a manufacturing context [29].

Romero et al. [25] established the relevance of adaptive manufacturing systems for operators with disability, making use of neuro-ergonomics to identify when an operator requires assistance and the system is triggered to aid the operator in their challenges. In agreement, Villani et al. [4] conceived a methodology for designing adaptive and human-oriented interfaces upholding the same human–machine symbiosis as per the works of [25]. At present, social sustainability within manufacturing envelopes the operator’s well-being and empowerment, as well as human-centric and adaptive design [78]. This should, however, emanate towards social sustainability involving diverse stakeholders (both their influence on the company, but also to foster an understanding of the company’s external implications on society [34]. Such considerations become even more avid when keeping in mind the broad employment base that is set by manufacturing companies globally, and also that, unfortunately, the role of humans in manufacturing is “too often overlooked” [34] (p. 689), especially when paralleled with the economic and environmental sphere of sustainability.

There is a dire inconsistency in knowledge when it comes to defining a multidisciplinary social impacts benchmark through measures, metrics, and indictors (MMIs) that are in accord [12,34]. This issue becomes even more blurred when combining socially sustainable employment of people with disabilities on the manufacturing shop floor, suggesting further room for improvement. An array of available guidelines, however, are still noted by Sutherland et al. [34], namely, the UN Global Social Compliance Program, Social Life Cycle Assessments (S-LCAs), and databases such as the Social Hotspot Database (SHDB).

The omnipresence of the core research gap, that is, the segregation of social sustainability and inclusion in relation to Industry 4.0 opportunities, is echoed in the works of Matt et al. [6], who conceptualise the notion of “socially sustainable urban factories” whose core lies in seeking skilled workers in manufacturing. In pursuit of social sustainability and cognitive awareness within the manufacturing shop floor, it is worth underlining current frameworks or methods that converge on the needs of the operator. Moore and William’s [52] social scaffolding algorithm envelopes the “emotional”, “tonal”, and “scaffold” scores [52] (p. 468). Positive emotions detected by the robot (such as happiness or excitement) are ranked a score that differs from that of a negative emotion (namely, apprehension or sadness), and the said score is analogous to the level of frustration and assistance that the operator would necessitate during the task. This scaffold score is set against a benchmark that determines when and how the social assistive robot (SAR) AIDA should interject. Other literary work has tapped into the exploration of social sustainable practices within manufacturing, such as the SO SMART project (2013–2015) [38], which aspired to set the ball rolling on social sustainability within European manufacturing industries. The project narrowed down on three levels of an ascribed “ecosystem”: the societal, the industrial, and the individual’s life balance.

Empowerment on the shop floor encompasses system adaptation to the worker’s needs, attention attributed to operator’s well-being, participatory design, and lifelong training and learning [2,9,10,12,29,40,48]. Apart from that, the social sustainable cap must be worn in a multitude of scenarios, commencing from the accommodation of operators with different abilities [43], but also in the comprehension of human–machine synergy. Lu et al. [48] investigated the layers of human–machine understanding, pointing out how the intended machine/system should be able to comprehend a human through instructions given and actions taken by the human. It is of utmost importance that the AT or the machinery employed on the inclusive workstation refrains from causing social disruption to quotidian processes, both to the user, but also to the other operators, and thus the machinery must be “fitting in the social and organizational structure” [47] (p. 4).

6.4. Key Takeaways from Section 6

The works introduced in this chapter depict that the notion of “sustainable Industry 4.0” is gaining rapid momentum. Notwithstanding, a clear rift persists between mere awareness of Industry 4.0’s relationship to sustainability, as opposed to properly implementing changes to sustain said relationship and its implications. In conjunction to providing an inclusive workstation with proper task allocation (social and economic sustainability) and achieving optimal efficiency at minimal emissions (economic and environmental efficiency), a wholly sustainable shop floor demands much more. It is only attained if the operators with disabilities are not siloed but work closely with other mainstream operators (avoidance of alienation). This weds together two of the core SDGs for this review: SDG 8 [82] (“Decent Work and Economic Growth”) and SDG 10 [83] (“Reduced Inequalities”), which are tapped into at a later stage of this review. The significance of appropriate control and autonomy is reiterated to empower the operator and grant a sense of ownership, epitomising social sustainable practices and closing the human–machine loop.

7. Identified Research Gaps and Future Work

This final segment of the review aims to envelope the major gaps identified throughout Section 4, Section 5 and Section 6, in relation to the fundamental research question posed in Section 1.

7.1. Main Research Question and the Main Outcomes of This Review

RQ. 

What is the current state of the art regarding the design of inclusive smart and sustainable workstations for operators with disabilities on the manufacturing shop floor?

The exhibited and critically reviewed body of work has identified research gaps worth elaborating on. Primarily, the majority of scenarios coupling people with disabilities, manufacturing tasks, and workstation design are based in sheltered workshops (such as in the works of [22,36,46,47,65,68]), thus failing to provide a clear image of the situation on actual manufacturing shop floors. A second major fissure in contemporary scholarly work is the tendency to converge to one type of disability rather than to acknowledge multiple disabilities as well as different skill sets and accommodations needed for the same task and setting. Accordingly, a potential approach to tackle these challenges would be to take a concurrent engineering perspective. Moreover, the integrated product development model [84] adapted by Francalanza et al. for Industry 4.0 [85] and Axiak et al. [86] for sustainability could therefore be adapted to take persons with disability as main stakeholders.

The existing literature may not clearly underline how individuals with a spectrum of disabilities (sensorial, mobility, and cognitive) would handle the same task simultaneously. To simplify this declaration, one may consider a noisy environment in which it is custom to wear ear protection. For an individual with a hearing impairment, this would be the ideal occurrence since ear protection inhibits the need to hear commands; yet, the same solution would pose a challenge for someone with a cognitive impairment who is easily startled by noise. Researchers sometimes fail to annotate such stochastic elements and subsequently propose solutions revolving around a generic notation that may benefit one individual at the expense of the other (for instance, research focusing solely on cognitively impaired workers or on operators with motor/mobility impairments, rather than operators with comorbidity).

Table A1. Summary of the major common topics elicited from the reviewed publications (in relation to the four disciplines) and the corresponding literature.

Four Disciplines	Subtopics Identified	Corresponding Literary Work References
Workstation design	Human skills in manufacturing	[2,4,5,6,7,9,10,12,18,19,20,25,29,33,34,38,40,42,43,45,48,49,50,51,52,55,58,59,61,64,66,71,78,90,91,92]
	Cognitive ergonomics	[4,5,6,10,11,19,20,29,34,37,39,43,48,50,61,63,64,66,69,71,74,75,76,77,80,90]
	Physical ergonomics	[2,4,5,6,7,19,20,25,29,37,45,48,49,61,63,66,69,70,80]
	User-centreed design/human-centred in manufacturing	[2,4,9,10,11,12,18,20,25,29,31,32,35,37,38,40,43,44,45,48,51,54,58,60,63,64,74,75,76,78,81]
	Universal design in manufacturing	[4,7,40,43]
	Participatory design in manufacturing	[7,10,18,34,38,47,62,68]
	Assembly line balancing and job allocation	[2,5,10,19,22,24,33,37,38,39,44,45,46,48,59,71,76]
	Inclusive workstation design for shop floor	[7,35,36,39,43,44,46,47,49,50,53,55,56,57,59,61,62,63,64,66,70,71,74,75,76,77,80,90]
Industry 4.0 and Industry 5.0	Industry 4.0 in manufacturing	[2,6,7,9,12,18,22,29,31,32,33,35,36,37,39,40,43,44,45,48,51,54,58,60,61,63,65,66,69,71,75,76,78,80,81]
	Automation in manufacturing	[2,4,6,7,9,10,12,20,25,29,31,33,37,43,47,51,58,63,66,71,74,75,78,90,92]
	I4.0 ATs for disabled operators	[4,11,18,19,20,22,33,35,36,37,39,40,42,43,44,47,50,51,53,54,58,59,60,61,62,63,64,65,66,69,71,74,75,77,80,90,91]
	I4.0 ATs for mainstream operators	[2,6,9,10,12,25,33,35,37,39,40,48,51,71,76,78]
	Human–robot collaboration	[18,19,29,37,44,47,48,54,58,59,62,63,65,68,80]
	Accessibility	[4,6,7,11,12,14,18,20,29,34,36,42,43,44,46,51,58,60,61,62,63,64,65,70,74,75,80,90]
	Industry 5.0 in manufacturing	[7,9,10,12,29,45,48,78,81,92]
Sustainability and social sustainability	Sustainability and Industry 4.0/5.0	[7,9,10,31,32,45,78,81]
	Socially sustainable manufacturing	[2,6,7,9,10,12,18,25,29,31,32,34,36,38,42,43,45,48,65,66,69,78,81,92]
	Social dimension of sustainability (generic)	[2,6,9,10,12,14,18,25,29,31,32,34,36,38,42,43,45,48,54,62,65,66,69,78,81,91,92]
	Economically sustainable manufacturing	[7,9,10,29,31,32,34,78,81]
	Economic sustainability (generic)	[9,10,29,31,32,34,78,81]
	Environmentally sustainable manufacturing	[9,10,29,31,32,34,42,78,81]
	Environmental sustainability (generic)	[9,10,29,31,32,34,38,42,78,81]
	Sustainable development goals (SDGs)	[7,32,34,42]
Disability in manufacturing and disability studies	Disability and Industry 4.0	[4,11,14,19,20,22,33,35,37,39,40,43,44,45,50,54,61,62,63,64,65,68,71,75,76,78,80,91]
	Disability in manufacturing	[4,5,6,7,11,18,19,20,21,22,23,24,33,34,35,36,37,39,40,41,42,43,44,45,46,47,49,50,51,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,74,75,76,77,78,80,90,91]
	Psychology and behaviour studies	[14,19,37,55,56,64]
	Universal design and disability	[7,18]
	Physical disability	[4,14,37,47,51,55,63,64,65,66,68,69,70,77,91]
	Sensory disability	[4,14,20,41,55,56,63,91]
	Cognitive disability	[4,11,14,20,22,35,36,37,39,41,44,47,50,51,53,54,55,56,57,58,60,61,63,64,65,66,68,74,76,91]
	Short-term memory loss	[53,65,76]
	Learning disability	[11,14,47,56,65,68]
	Visual	[42,51,53,55,56,57,63,65,68]
	Disability in employment and accommodations	[4,7,11,14,19,20,21,22,37,40,42,43,44,46,47,49,50,51,53,55,57,59,61,63,66,68,69,70,71,75,80]
	Sheltered workshops	[22,36,47,60,64,65,66,68,74,75,76,77]

in Appendix A avidly showcases this element of segregation.

On the other hand, the mapping of social sustainability within a field currently dominated by techno-centricity may also give rise to unprecedented considerations related to socio-ethical issues [10,43], data security, and the risk of privacy invasion [10,19] (by, for example, having a robot that monitors the human to be able to adjust to the operator’s needs). Other loopholes include lack of trust (or over trust) between humans and machines, which becomes even more relevant when the operator has a cognitive disability and is not able to anticipate the robot’s moves [37]. Such issues pose a dichotomy—is technology used to relieve mental stress or is it incurring additional stress, anxiety, and scepticism due to higher expectations and steeper learning curves?

Within this review’s pursuit of understanding workstation design for people with disabilities, a common denominator that emerged was the technological equipment chosen—touchscreens or AR to present instructions, over-head cameras to detect the input-output of parts from their containers, and sensors to detect any changes in the human’s behaviour over time. However, despite such a strong trend, none of the reviewed work allude to environmentally sustainable choices made throughout the design of the workstation (such practices could have included designing for manufacturing and assembly (DfMA), applying life cycle costing to the technologies employed or adopting design for sustainability practices).

On another note, further patterns have also been recognised, such as the recurrence of instruction reinforcement through prompts and visual cues, reward systems, and positive remarks when the task is properly executed. Most of the research portfolio that ties operators with disabilities to manufacturing revolves around tailored, personalised, and adaptive instructions, as opposed to the actual workstation space and design. On the one hand, every operator is a learner in their own right and follows a unique pedagogical journey to comprehend how, say, a new task is to be executed. This upholds instruction delivery as crucial especially for individuals with disabilities, yet in doing so, academia fails to weave in other aspects of workstation design that are vivid in manufacturing [49]. Research within this remit should thus point towards accommodating for ergonomics after multimodal adjustments are made, understanding how to cater for fixed production times and quotas, and in general, what the role of an inclusive workstation entails for assembly line balancing [46]. Furthermore, a number of case studies examined (both those occurring in sheltered workshops and on manufacturing shop floors) have only conducted an evaluation based on a one-time exercise or after just one sitting. In fact, more studies should follow in the footsteps of Funk et al. [33] and are advised to opt for long-term observation. This ensures a holistic depiction of any changes in the user’s behaviour and a true reflection of the challenges or amendments that might be requested after a prolonged period of use [11,33]. Such approaches secure a robust and resilient case study under the umbrella of Industry 5.0. Future work in this multidisciplinary dome should prompt enhanced design and testing of technologies for employees that steer away from the status quo, keeping in mind that user-centred workstation development and human–robot collaboration should not fail to also cater for disabled operators without replacing them. An influx of exoskeletons, augmented and virtual (AR and VR) reality multimodalities, and collaborative robotic arms will define the workspace of the future.

In conclusion, Figure 9 adapts Giddings et al.’s [87] nested sustainable development diagram to convey the dependencies of the areas tapped in this work. This figure answers the study’s research question and uncovers ground for future work, as is elaborated in the next section.

7.2. A Bright Future for Inclusivity on the Shop Floor in the Age of Industry 5.0: Future Work

This review has contributed a contemporary snapshot of the trends, patterns, and practices that surround workstation design for people with disabilities on the shop floor, as well as the influence of smart technology and sustainability. Despite the uncertainties elicited through the findings, there is ample room for improvement. Effort should gear towards merging all the four disciplines introduced in Section 1 into one, taking-off with the notion that each operator is different and that only with the involvement of people with disabilities through all stages of design shall a true inclusive workstation come through. Thus, future work should commence with establishing collaborations between a triad of stakeholders: academics, persons with disabilities, and industry practitioners, in order to devise workstation designs that exceed both the industry’s and the individual’s needs. Priority should also be appointed to operators with disabilities and their experience on the shop floor, thus prompting timely smart assistance and modifications where and when needed. As detailed in the review, enabling technologies such as cobots, AR and VR, and multi-modal control have opened a window of possibilities within the remit of inclusive employment and assistive workstation design. Despite the potentials and benefits promised by such technologies, future research should also foster a social and humanitarian outlook, considering secondary challenges that may arise from these technologies over time. Such encounters could include the relationship established between the operator and the AT (especially if the operator feels that using the AT is compulsory, yielding an adverse relationship between the dyad); issues such as over- or under-trust, or over- or under-load (both cognitive and physical); and unprecedented system breakdown and its effect on the operator, among other issues.

Future work should not merely steer in such a direction (onboarding ATs) but should venture to assess the learning progress, well-being, and productivity of operators with disability over time. This would ensure that future workstation design weaves these lessons learnt and permits operators with disabilities (both present and especially future) to have a say from the onset of planning. In this approach, workstation modifications can be made with ease. Ultimately, this fuels social sustainability, providing equal opportunities without prejudice, as urged by the sustainable development goals.

Literature on sustainability within Industry 4.0 exists, yet most of the work tends to shed light on one or two of the three sustainability pillars (which are economic, environmental, and social). This becomes even more evident when inclusion of persons with disabilities is added to the equation, since one may argue that designing for inclusivity (hence addressing the social pillar) might incur a high financial and environmental cost. This should be averted, as catering for one pillar should not come at the “expense” of the other two. Akin to this, in Kyoto, Japan, OMRON Taiyo is the epitome of a socially sustainable manufacturing company, with more than half of its employees having either a physical, intellectual, or mental disability [88,89]. OMRON Taiyo’s inclusive workstations utilise Andon boards, as well as coloured floor markings, to facilitate visual accessibility with minimal need for word interpretation. Such a case study should act as a prompt for academics to develop well-needed practical solutions that would be fruitful for both operators with disabilities (accessible employment) and companies alike (productivity and feasibility). Following this industrial insight, Section 7.3 provides a theoretical roadmap that spurs academics towards future work that upholds sustainability, an area that was seldom explored within the literary context of inclusive design.

7.3. Shifting the Direction towards a Holistically Sustainable Design Approach: Adopting the SDGs for Future Work

It is equally crucial for upcoming researchers to consult with and enmesh the sustainable development goals within smart and inclusive workstation design, a combination seldomly annotated in prior work (as is the case within this review) yet exhorted by Industry 5.0 researchers such as [29]. Engineering researchers are therefore invited to consult with four major relevant (yet very distinct) SDGs within the umbrella topic proposed (to encapsulate all four disciplines):

-

SDG 8—Decent Work and Economic Growth (primarily addressing Disciplines 2 and 4).

-

SDG 9—Industry, Innovation, and Infrastructure (primarily addressing Disciplines 1, 3 and 4; also recommended by [29] in pursuit of adopting SDGs towards Industry 5.0.).

-

SDG 10—Reduce inequality within and among countries (primarily addressing Disciplines 2 and 4).

-

SDG 12—Responsible Consumption and Production (primarily addressing Disciplines 3 and 4; also recommended by [29] in pursuit of adopting SDGs towards Industry 5.0.).

Such an unforeseen amalgamation shall be undertaken and exemplified by the authors in future work within the remit of Industry 5.0-oriented inclusive workstations, as depicted in Figure 10.

8. Conclusions

The emergent human–machine symbiosis has accelerated the need for future research to orient its directions toward workforce diversity, predominantly the collection of industrial examples and in situ case studies with all the intended user groups. This research work has contributed an in-depth PRISMA review surrounding the design of smart, sustainable, and inclusive workstations. To the authors’ best knowledge, this is the first review that bridges four segregated disciplines (workstation design, Industry 5.0, sustainability, and disability) into unification, thus aspiring to act as a reference point for researchers, engineers, employers, and managers alike.

This thorough review also attempted to debunk the impression that people with disabilities are not apt to work on an assembly line and execute diverse manufacturing tasks. As highlighted in much of the work analysed, with the proper design considerations and suitable technologies, the opportunities on the shop floor can be furthered within the context of Industry 5.0. The most evident obstacle in this multidisciplinary topic may point towards academia and industry not always paralleling each other and failing to practically implement what has been theoretically ascertained. The literature may acknowledge the needs and potent solutions, such as those related to social sustainability, yet industry fails to implement them. This will indeed be the authors’ focus in upcoming research work, that is, unveiling current hurdles that obstruct operators with disabilities from working on the industrial shop floor and implementing suggestions from an industrial perspective. Despite the advantageous possibilities that emerge on the surface, the pillars that sustain Industry 5.0 may give rise to unforeseen challenges which will demand an amalgamation of multiple disciplines to balance out the three fundamental requirements (resilience, human-centricity, and sustainability). This is evidently already challenging in theory, let alone when applied in practice, leaving room for more crossovers between academia, industry, and society.