A Participatory Content Authoring Workflow for Augmented Reality at Industrial Maintenance

Abstract

The adoption of augmented reality-based instructions enhances maintenance operations by shortening job completion time and reducing errors. However, scaling augmented reality in industrial settings remains costly since content authoring demands computational skills such as 3D modeling and programming. Furthermore, processes can easily become obsolete, causing maintainers to abandon written instructions. So, we propose an augmented reality-based participatory content authoring workflow for maintenance tasks. We followed the Design Science Research paradigm, which included a literature review, the conception of a workflow, and a simulation to evaluate the proposed workflow’s validity. We found that current workflows overlook participatory content authoring involving maintainers and that most research focuses on describing the technical architecture of proposed systems rather than a workflow that supports the use of technology in industrial settings. Regarding our proposed participatory workflow, most respondents stated it was simple to use, improved their capacity to develop augmented reality content and would help the industry adopt augmented reality. As a result, our participatory authoring workflow can optimize augmented reality content authoring during maintenance, encouraging the maintainers’ interaction, and provide opportunities for procedure improvement. We conclude that non-programmer-friendly augmented reality software tools save content production time while enhancing users’ perceptions of their own technological talents.

1. Introduction

Maintenance is critical in industrial organizations, boosting competitiveness, and contributing to long-term development [1]. Maintenance operations are knowledge-intensive for maintainers because they frequently involve massive amounts of equipment, subsystems, components, and high-complexity tasks spanning from machine diagnostics to repair [2].

Although valuable, knowledge acquired in an organization can soon become obsolete, whether owing to the replacement of knowledgeable professionals with less experienced personnel, poorly standardized activities, technology obsolescence, insufficient record keeping, and loss of processes [3]. Furthermore, rising industry competitiveness and fast technology advancements have resulted in the requirement for each employee to operate more efficiently and effectively, making information readily communicated between experts and novices [4]. To achieve these requirements, knowledge management plays a vital role in establishing systematic methods of identifying, acquiring, and deploying an organization’s intellectual capital [5]. As a result, the knowledge transfer process, described as “the transportation of knowledge from one place, person, system, or property to another” [6], is critical for the maintenance’s long-term viability.

Companies that produce and disseminate knowledge throughout the organization, integrating it into new technologies and products, are successful [7]. This knowledge might be tacit (learned by experience over time, very personal, and difficult to communicate) or explicit (expressed in words, and shared in the form of data, visual, sound, product specifications, or manuals). When tacit knowledge becomes explicit knowledge, a transfer occurs an “externalization” [7]. In maintenance, this is frequently accomplished through the development of processes.

A maintenance process is a series of operations carried out by one or more specialists in order to keep an item’s dependability. The method comprises descriptive text and may incorporate diagrams and illustrations to help maintainers comprehend the material. Although the instructions were originally printed, it is now usual to obtain them electronically [8].

The adoption of augmented reality (AR) procedures in maintenance has been found to save time and reduce mistake rates. AR technology supplements the real world with virtual objects (generated by a computer) that appear to coexist in the same space as the real world [9]. Nevertheless, the application and scalability of AR in an industrial environment continue to have a high development cost, particularly in its content authoring process, because numerous computational skills such as modeling, programming, and animation are still required.

Furthermore, present procedures (electronic or printed) are frequently perceived as irrelevant or outdated by maintainers since they do not reflect the reality of the operation [10]. Change is a constant in industrial processes, forcing the ongoing modification or updating of one or more procedures [11]. Most of the time, these procedures are written or revised by specialists with limited actual operating expertise, resulting in a gap between knowledge and industrial reality [12]. Additionally, procedures are rarely required to accomplish an activity [11]. Most of the time, maintainers know what to do and can adapt to changing conditions. As a result, the efforts necessary to create procedures are often given low priority, and their deployment is delayed. Even when procedures are accurately defined, updating them is often given low priority, resulting in obsolete knowledge and a loss of confidence among operators and maintainers.

As an example, in a study with 400 maintenance technicians and managers from industry, the interviewees were asked to propose improvements to contribute to greater use of the procedures, and the most frequent suggestion was “to involve end users in the development” [12], corroborating that the participation of the maintainers is critical for the successful adoption of the procedures by its users [11].

Thus, while conventional generating maintenance procedures generally do not involve user participation and result in outdated and irrelevant processes that are poorly accepted by users, augmented reality may improve maintenance procedure efficiency, but AR-based content development still necessitates computational abilities such as modeling, programming, and animation [13].

In this context, we propose a participative content authoring workflow for maintenance tasks based on augmented reality.

This paper is organized as follows: Section 2 describes the materials and methods used, Section 3 describes our findings, and Section 4 presents our conclusions and suggestions for further investigation.

2. Materials and method

We adopted the Design Science Research (DSR) paradigm. In addition to a knowledge contribution, DSR contributes to the real-world application environment from which the research problem or opportunity is drawn [14].

Our method parallels that described by Gregor and Hevner [14], which includes six steps: (1) identify the problem; (2) define solution objectives; (3) design and development; (4) demonstration; (5) evaluation; and (6) communication.

In steps 1 and 2, we realized the problem situation and defined the solution objectives, which is to develop an augmented reality-based participatory content authoring workflow for maintenance tasks.

In step 3, a systematic literature review was designed. Initially, we conducted a preliminary examination of 10 works addressing the usage of augmented reality and the efficiency of maintenance processes [15]. Then, this preliminary approach was improved using the strategy provided in Ref. [16]. The phrase “((framework OR workflow) AND (augmented reality OR mixed reality OR extended reality) AND maintenance)” was searched in the databases IEEE Xplore, ScienceDirect, and Scopus to answer two research questions: “Q1: Are there any workflows proposed for augmented reality authoring content that allow participatory content creation among end-users?” and “Q2: What are the differences and similarities between the workflows encountered?” The papers were then screened and criteria were applied to include works of type articles, magazines, and conferences published since 2013 and excluded works that were not applicable for industrial maintenance and did not provide an augmented reality workflow. We found nine papers that were categorized by the authorship type indicated by Ref. [13]. Table 1 shows the results of this strategy.

Ref.	Authoring Type	Participatory Authoring?	KPI
[17]	By annotations	YES	N/A
[18]	By annotations/manual	YES	Time
[19]	By annotations	YES	N/A
[20]	By boxes	NO	Time and Error Rate
[21]	Manual	NO	N/A
[22]	By annotations/manual	NO	N/A
[23]	Automated	NO	N/A
[24]	Manual	NO	N/A
[25]	Manual	NO	N/A

Table 1.

List of retrieved works.

In step 4, in light of a typical authoring workflow used in the manufacturing industry, we demonstrate a proof-of-concept of our proposed workflow of participatory content authoring in AR at maintenance, tested and revised it through expert reviews, and exposed preliminary versions to researchers in seminars and workshops, including International Symposium on Innovation and Technology (SIINTEC) and Brazilian Maintenance and Asset Management Association Conference (ABRAMAN).

In step 5, we evaluated the proposed workflow in terms of validity criteria, since DSR includes gathering evidence that the artifact is useful, meaning that the artifact works and does what it is intended to do [14]. We evaluated our workflows’ validity by carrying out a simulation with 10 manufacturing specialists divided into three teams based on their similar skills. The age group of the participants was 50% between 20 and 30 years old, and 50% between 31 and 48 years old; eight participants had skill and interest in new technologies; eight participants were experienced with the use of augmented reality in maintenance; and seven participants had experience in creating procedures, despite the fact that only two participants considered themselves capable of creating AR maintenance procedures. The maintenance procedure simulation was performed on a Cessna C500 aircraft (Figure 1) and consisted of five stages.

Figure 1.

Aircraft C500 Cessna subjected to the simulation.

At Stage 1: ar creation and visualization tool training, a 60-minute tutorial on the AR content authoring tool was given to the 10 participants. The Viscopic Pins software was utilized, which was designed for non-programmers to create AR procedures. The participants were shown how to produce AR content in the tool and the different forms of 2D (Figure 2a) and 3D (Figure 2b) content visualization. Following the demonstrations, participants put the authoring tool to the test and developed original content based on what they had learnt. Each team had 10 minutes to utilize the tool. A 3D CAD drawing, images, and videos were made accessible for use in the procedure’s development.

Figure 2.

2D (a) and 3D (b) content visualization.

In Stage 2: aircraft inspection training, the participants received a 30-minute seminar delivered by an aircraft maintenance expert, with the knowledge relevant to the simulation to be performed.

The Aircraft inspection simulation with AR procedure took held at Stage 3: aircraft inspection simulation with AR procedure, with the groups being prompted to perform three macro inspections: oil check, attack board fuselage, and supply valve. The precise location of where the inspections should be performed was included in the AR method and followed the route depicted in Figure 3a.

Figure 3.

Maintenance simulation route (a) and oil check stage (b).

Because the stages did not follow an ordered flow, the participants suggested that the sequence of inspections be changed as a chance for improvement. In addition to completing the checklist, if any of the stages were not completed within the time frame set, the participants would be required to conduct some action to rectify the situation. During the oil check stage, the oil level in each turbine (represented by a plastic bottle of water, as shown in Figure 3b) had to be checked and refilled as needed. The amount of oil required was not specified in the AR method. This information was included in the materials given to the crew on a clipboard and on a page near the warehouse. The turbines required 30 liters of oil, however, the left turbine had only 15 liters. The oil container was in the warehouse, ready for the teams to fill. A picture in the attack board’s fuselage revealed the exact spot (marked in red) where the crew should count the missing rivets and place them if necessary. The rivets were accessible at the warehouse, and it was essential to replace the 16 rivets that were missing. Furthermore, the complete 3D leading edge appeared overlaid in real time with AR (Figure 4).

Figure 4.

AR Fuselage stage.

Finally, at the supply valve stage (Figure 5), the groups were required to seal the valve cover by softly tapping it (information supplied verbally during training), in addition to locking it. In augmented reality, a snapshot of the closed valve was made available.

Figure 5.

AR supply valve stage.

The groups developed a competitive environment. The procedure’s time and mistakes were recorded. A minute was added to the overall inspection time for each error committed by the group. If the instructions supplied to the task competition were insufficient, the groups might contact the aircraft expert for assistance, but each assistance added 1 minute to the task completion time.

The participants examined what information might be added to the AR procedure at Stage 4: update of AR procedure, guaranteeing that the instructions would be full without the usage of paper or oral directions. Each group was given 1 minute to shoot photos or videos.

At Stage 5: aircraft inspection simulation with updated AR procedure, the improvements recommendations were reviewed to develop an ideal procedure, and the participants engaged in the aircraft Inspection collectively to evaluate the result. Participants answered a post-dynamics questionnaire at the end of the simulation.

Step 6 of the Design Science Research approach entails communicating our findings in this work.

3. Results and discussion

In the sections that follow, we describe our findings from the literature review, the development of the proposed workflow of participatory content authoring in AR at maintenance and the simulation to evaluate it.

3.1 Participatory content authoring workflows for AR at maintenance

The systematic literature review yielded nine works, which are listed in Table 1. These papers were categorized based on the authoring type specified by Ref. [13], whether the addressed workflow suggested participatory creation with end-users, and whether any Key Performance Indicators (KPI) were used to demonstrate the efficacy of the workflow.

Refs. [17, 18, 19, 22] addressed workflows that allow the interaction of users through virtual annotations or block edits of the system, allowing participative content authoring.

The solution developed by [17] allowed maintenance technicians to edit the positioning and content of information in AR using a mobile application. The technician’s edition was sent to the content developer to validate the information and make it available on the system, naming this type of authoring “bi-directional.” Content authoring was also allowed online for technicians who were doing remote maintenance, assisted by another technician who could create the content in AR and make it available in real time to another maintainer.

The system proposed by Geun-Sik et al. [18] was developed in the aviation industry and consisted of three modules: augmented reality (AR) module; knowledge-based system module (KBS); and a unified platform with an integrated UI/UX module between AR and KBS. At the time of maintenance, the system recognized the component and made the most relevant information for that activity on the screen using artificial intelligence (AI). Despite the use of AI, the manuals in the KBS module were made manually by specialists in the field. Within the tool, it was possible to include notes from users to specialists.

Holger et al. [19] developed an AR-based approach using virtual notes as a communication platform between maintainers. These notes can be attached to components without predefined markers, eliminating the need to configure the AR system. In addition to textual information, these notes can refer to different interaction methods, such as factory process data, manuals, audio, video, and so on.

Some studies did not propose participative authoring. Lamberti et al. [20] presented the structure of the software developed in the EASE-R3 project, which explores reconfigurable AR procedures and remote assistance to overcome some of the limitations of current solutions. Michael et al. [21] proposed a system to improve communication between maintainers in the workplace. Despite providing real-time data from the machine, the system does not propose instructions on how to carry out an activity. However, it provides the status of the activity carried out by a maintainer to notify other interested parties when they can proceed with other activities.

Dimitris et al. [23] presented the development of an augmented reality application for a CNC bending machine. From the equipment’s CAD information, the system was able to create the necessary instructions for configuring the machine, as well as make available the 3D step-by-step that already existed before the application. A feature was also incorporated into the system that made it possible for the maintainers to communicate with engineers, effectively addressing possible problems in production.

Siew et al. [24] proposed the ARMS system, which provides maintenance support by integrating three modules: visualization, haptic (tactile), and detection based on markers. The developed system provides feedback to users regarding the location of maintenance using tactile sensors. In addition, the content available to the user is different depending on their level of knowledge. This function is automatically changed if one of the stages to be completed by the maintainer takes longer than expected or manually by the user himself.

In the workflow proposed by Havard et al. [22], four departments were approached: design, information technology (IT), specialists, and operation. The solution mixes RA and RV in the flow, and each department has a well-defined responsibility. The design develops the 3D model of the system; TI creates the AR and VR library; the specialists develop the interactions; and users, if they are performing a task for the first time, undergo VR training first and then proceed to do the task with AR assistance. At the end of the activity, the user can give feedback to the specialists who will define the need for updates. Despite not proposing a maintenance KPI to validate the workflow, the authors reported a first analysis indicating that a developer takes 96 minutes to create an operation stage in RA without the solution. With the solution, this process decreases to 51 minutes.

Zubizarreta et al. [25] proposed a model composed of a set of software libraries and two main applications: an authoring tool that creates a database with information on the steps to be performed, including several types of multimedia information (text, video, VR and AR animations) and a guidance tool that presents that information using a mobile device. Both tools are designed to provide a simple user interface. The user can access a machine’s process information and navigate through each stage of a maintenance operation. An interface also allows the creation of new content (for the authoring tool) or its visualization (using the guidance tool). Although tracking works with markers or based on the model, importing the equipment’s geometry is necessary for the system to recognize the positioning of the content. The study focuses on the technical approach of image recognition, and the tests carried out measured indicators of system performance and not its impacts on users’ efficiency.

The majority of addressed workflows [17, 20, 21, 22, 23, 24, 25] do not consider the participative authoring of augmented reality content involving their end-users, in this context, the maintainers. Most studies are concerned with describing technical details about the architecture of the proposed systems instead of a workflow that facilitates the use of technology in an industrial environment. In addition, except for Ref. [18, 20], all the others did not use KPIs to measure the impacts of their workflows on maintenance. Among the proposals that allow the participation of end-users in the development of procedures, only Ref. [17] developed a workflow with the interactions between the stakeholders of the process (developer and user), with the rest presenting a model of the system architecture.

As a result, there is a potential to design a workflow that, in addition to allowing non-programmers to author content, involves the participation of its end users, therefore contributing to the industry’s adoption of AR.

3.2 Proposition of a participatory content authoring workflow for AR at maintenance

Maintenance procedures are typically developed by a professional with minimal practical experience, resulting in a detachment from industrial realities [12]. Furthermore, the process of developing a procedure requires a significant amount of time from the specialists concerned, which might jeopardize its updates [11]. Figure 6 depicts how procedures are frequently generated in the manufacturing industry, exemplifying the procedure creation cycle outlined in Ref. [11].

Figure 6.

A typical workflow for creating maintenance procedures in the manufacturing industry.

To start, the maintenance engineer is in charge of creating the digital procedure in a text editor. The Quality Engineering sector makes a template with the company’s internal regulations available on the document management system (DMS). The procedure’s content will be determined by the maintenance engineer’s knowledge. He will be able to apply his tacit knowledge, arrange meetings with maintainers and specialists, and review manufacturer’s instructions. Photos can also be included by the maintenance engineer to help explain the methods outlined. The maintenance engineer prepares the procedure and uploads it to the DMS, where it is reviewed by the quality engineer. It is crucial to note that this procedure does not occur for all machine faults, but just for those that the maintenance engineers judge to be the most critical.

The quality engineer then verifies that the procedure is clear, adheres to safety and quality requirements, and follows the company’s internal norms. If something does not meet the standard, the document is returned to the maintenance engineer until the problem is resolved. After the quality engineer has authorized the procedure, it is required to contact the training department before it is made available in the system to all employees.

Then, the trainers in each area are responsible for carrying out training for that procedure and collecting physical signatures from each trained maintainer. In many instances, the department itself creates a presentation with videos, so that the information is understood more clearly, however, this material is not available for later consultation at the DMS. In addition, there is no urgency to train maintainers, as standard operating procedures, that is, instructions for operators who work directly with the product, are prioritized to maintain the quality and safety of the process.

The maintenance engineer will then make the document available on the DMS, and the quality engineer will print physical copies.

Finally, the maintenance engineer removes the printed procedures from the quality department and places them in a physical folder at the area where the activity occurs. This stage can take a long time for the maintenance engineer depending on the number of areas that will employ these procedures, and it is distant from the reality of the maintainers, who seldom use physical manuals.

In the meanwhile, when a maintainer must do a maintenance task, he checks to see if the procedure is accessible in the physical files or the DMS. If the procedure is discovered, the activity is carried out with the assistance of the instructions created; if the procedure is not discovered, the activity is carried out based on their knowledge.

The availability of current procedures requires time due to workflow constraints. Because it is dependent on training and signature collection prior to disclosure, and because printing physical copies generates rework for engineers who have already made digital copies available, maintainers are only involved if the maintenance engineer chooses to request assistance in the creation of the instructions; otherwise, there is no direct feedback from the maintainer to the maintenance engineer.

In this work, we propose a participatory content authoring workflow for AR at maintenance (Figure 7), based on the workflow described in Ref. [17] and taking into account the limitations of the procedure creation pipeline in the manufacturing industry.

Figure 7.

Our proposed participatory content authoring workflow for AR at maintenance.

As part of the workflow, an authoring tool that allows any user to produce AR content without programming is required.

Our workflow takes into account just the end user (maintainer), a specialist (maintenance engineer) who will assess the approach the information was created, adding more data as needed, and an auditor (a quality engineer) to ensure that the procedure adheres to the company’s rules.

The training department was eliminated since it is considered that AR technology provides the end user with the essential comprehension of the process.

The user will first open the AR application and search for the procedure to begin. There are two options from there:

• If the procedure cannot be found, the user must complete the task without the aid of AR.

• When the activity is over, the user creates an on-site procedure (with images, videos, and/or text components), saves it, and exits the app.

• If the procedure is found, the user accomplishes the activity with AR assistance. At the conclusion of the activity, the user is asked if the task was accomplished simply by following the instructions. if the response is:

  • No: An opportunity for improvement was discovered. The user must alter or add the portion in question, save, and exit the application.

  • Yes, simply exit the application.

• When receiving updates or new procedures, the specialist determines if the content created requires more information, sends it to the auditor to conduct the conference in accordance with the company’s requirements, and makes it available in the system, validating the creation.

3.3 Proposed workflow test simulation

The five stages of the simulation were completed by the three groups: augmented reality creation and visualization tool training, aircraft inspection training, aircraft inspection simulation with AR procedure, participant update of AR procedure, and aircraft inspection simulation with updated AR procedure.

Table 2 shows the improvement suggestions offered by each group.

	Group 1	Group 2	Group 3
Change stage	x
Identify the oil level in the turbines	x	x	x
Add valve closure video	x		x
Describe Turbine Position	x	x
Describe valve side		x
Identify required security items			x
Check the integrity of rivets			x
Check valve integrity			x
Verify signal light			x
Request date and signature at the end of the procedure			x
Total of improvement suggestions	4	3	7

Table 2.

Suggestions for improvement from groups.

In general, the user training time allowed all groups to propose improvements. Most improvements were offered by Group 3, whose participants had prior expertise with procedures and AR authoring. Their suggestions went above and beyond what was required to examine in the checklist (for example, check signaling light), as well as the inclusion of an alert on the personal protective equipment (PPE) required to carry out that inspection. We infer that because the group has more expertise in creating procedures, the improvements emerge in a more intuitive manner than the other groups.

The only group to suggest a modification in the order of stages was group 1 (Figure 8), which was formed by participants who had no experience in creating procedures.

Figure 8.

New route proposed by group 1.

As noted, the lack of order, following the inspection, was included on purpose in the initial design of the activity in order to force the teams to go further than required. In other words, this improvement idea directly contributes to reducing total inspection time.

All groups argued that the AR procedure previously informed them of the right amount of oil to be poured into each of the plane’s turbines (avoiding the use of paper). Two groups proposed, including greater information on the plane’s left and right sides to improve procedure accuracy.

Each group made at least one suggestion that the others did not, highlighting the benefit of this workflow. By enabling any user to offer improvements, the procedure may be enriched in a variety of ways, leading to continuous updating and detailing.

All groups had seen the location of pins as a source of concern during the simulation. Because all groups chose 2D visualization (which displays the information on the side of the device rather than in the actual world), the groups read the information but believed they were in the proper place rather than following the lines that went to the pin. Group 2 was the most influenced by pin location since they spent more time doing the inspection simulation. As a result, it was the only group to argue that the valve location was previously given as an instruction in the RA procedure.

In terms of errors, group 1 placed the rivets on the back of the plane, entirely overlooking the location of the repair. Groups 2 and 3 just listed the amount of rivets that were missing in the right section of the aircraft on the checklist, but they did not travel to the warehouse to make the repair. These two groups likewise failed to load the aircraft with oil and were unable to locate the necessary information on the amount required. Groups 2 and 3 complained in a post-simulation conversation that it was unclear that when a deviation was discovered, it was important to effectively repair it rather than simply writing it down in the checklist.

Although Group completed the exercise in less time, it may be assumed that if the team had really completed the necessary aircraft repairs, the time would have been greater. This illustrates that, while evaluating an AR application, it is critical to examine the measurement of errors in the procedure in addition to studying the effects of technology over the course of an activity.

The groups were permitted to try the improved instructions and were instructed to use the 3D view after updating the procedure with all of the requested improvements.

At first, the participants were upset with a lack of precise step-by-step instructions in AR, leading many users to distrust the technology’s ability to aid with maintenance operations. On the other hand, because all of the recommended improvements were implemented, the procedure grew more comprehensive, surprising the participants with its ability to reduce downtime.

Following the simulation, participants were asked to complete a survey on their perceptions. The key findings were:

• The final improved procedure was chosen by nine participants as the best learning method for accomplishing the aircraft inspection. Only one participant chose the aircraft specialist’s spoken instructions.

• Six participants agreed that the task could only be completed by following the AR instructions, whereas four disagreed.

• Eight participants increased their view of being capable of creating AR procedures by at least one level on the Likert Scale.

• In AR, nine participants selected films as a visualization approach.

• According to five participants, one of the most challenging aspects of completing the inspection was a lack of knowledge of AR visualization. The greatest obstacle, according to four, was a lack of technical understanding of an airplane.

• Eight participants feel that our proposed workflow will make AR deployment easier in industry.

• Five participants said the workflow was simple to follow.

The findings demonstrate that following the proposed workflow, the participants felt more capable of creating AR content; yet, the lack of expertise in visualizing the authoring tool in AR was one of the interviewees’ most challenging issues. More extensive training on the instrument is required in this scenario to overcome these challenges and ensure the participants’ knowledge of the procedure.

Most respondents said our proposed workflow was straightforward to implement, boosted their ability to create AR content, and would contribute to the industry’s adoption of AR.

As a result, our proposed participatory authoring workflow can help to optimize the authoring of content in AR at maintenance, promote end-user participation, and offer additional chances for procedure improvement.

4. Conclusion

There has been limited investigation and development of participatory workflows that involve users in procedure creation. Users’ participation boosts the acceptability of the created instructions by introducing the realism of their work in the form of procedures. End users benefit from participatory creation when incorporating AR by simplifying the procedure creation process and increasing the possibilities of this technology functioning as maintenance support in their daily operations.

In terms of contributions, most research on augmented reality in maintenance focuses on technical aspects without considering the existing maintenance environment in which this technology would be used.

This study demonstrated how, from the start of implementing these systems, user participation is relevant to ensure procedure updates and raise the likelihood of adherence to usage. Otherwise, AR procedures, like digital or physical procedures, will become outdated. Identifying these gaps before adopting new technology helps increase the likelihood that these implementations will be successful.

Further study might widen the simulation to encompass a greater number of individuals, as well as studies with statistically significant samples and industrial settings. We also propose that future studies investigate how to overcome the technological limitations of AR; analyze how automated authoring would influence the proposed workflow; and analyze users’ adherence to using the procedure.

Acknowledgments

The authors acknowledge the National Council for Scientific and Technical Development (CNPq) for financial support; IW is a CNPq technological development fellow (Proc. 308783/2020-4).

Conflict of interest

The authors declare no conflict of interest.