A Systematic Literature Review on Serious Games Methodologies for Training in the Mining Sector

Abstract

High-risk industries like mining must address occupational safety to reduce accidents and fatalities. Training through role-playing, simulations, and Serious Games (SGs) can reduce occupational risks. This study aims to conduct a systematic literature review (SLR) on SG methodologies for the mining sector. This review was based on a methodology inspired by the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. Three research questions were formulated to explore how SGs contribute to immediate feedback, brain stimulation, and training for high-risk scenarios. The review initially identified 1987 studies, which were reduced to 30 relevant publications following a three-phase process: (1) A search string based on three research questions was defined and applied to databases. (2) Publications were filtered by title and abstract. (3) A full-text reading was conducted to select relevant publications. The SLR showed SG development methodologies with structured processes that are adaptable to any case study. Additionally, it was found that Virtual Reality, despite its implementation costs, is the most used technology for safety training, inspection, and operation of heavy machinery. The first conclusion of this SLR indicates the lack of methodologies for the development of SG for training in the mining field, and the relevance of carrying out specific methodological studies in this field. Additionally, the main findings obtained from this SLR are the following: (1) Modeling languages (e.g., GML and UML) and metamodeling are important in SG development. (2) SG is a significant mechanism for cooperative and participative learning strategies. (3) Virtual Reality technology is widely used in safe virtual environments for mining training. (4) There is a need for methodologies that integrate the specification of cognitive functions with the affective part of the users for SGs suitable for learning environments. Finally, this review highlights critical gaps in current research and underscores the need for more integrative approaches to SG development.

1. Introduction

Mining is a high-risk industry, and all miners, from heavy equipment operators to routine mine workers, require safety training. Therefore, continuous safety training is crucial in this sector, as demonstrated by Zujovic et al.’s study [1], and helps companies maintain a good safety record, as shown in research by van Wyk and de Villiers [2]. Training allows for improving the mining industry’s production processes and competing in the global market [2]. However, one of the most common problems in this sector is providing safety training [3].

In turn, Serious Games (SGs) have been defined as digital games whose purpose is not only to entertain the player but to go beyond entertainment [4]. In [5], the definition of SGs from [6] is adopted as mental contests in computers that, through entertainment, promote education in areas such as government, health, and public policy. From a more structural perspective, ref. [7] proposed that an SG is represented as the sum of utilitarian functions (training, message dissemination, or data collection) and a video game. SGs seek to achieve specific objectives such as education, training, simulation, or health promotion [8]. Simulation is one of the most widely used genres [9] and has been tested in different contexts to strengthen experiential learning. SGs are digital simulations similar to video games that combine engagement, reward, and enjoyment with educational and training purposes, utilizing gamification elements for enhanced training applications [10,11]. Gamification has gained traction in non-game systems and SGs to motivate users toward specific behavioral or psychological outcomes, such as faster learning or profile completion [12,13]. Gamification in learning environments can positively affect physical, cognitive, emotional, and social well-being, mitigating health risks and improving learner engagement [14]. Learner engagement depends on his/her motivation and emotional state [15]. The relevance of emotions in education has led to the use of affective recommendation systems in learning environments [16] and the need to analyze the feelings of learning resources based on their textual content [17].

On the other hand, the research community has explored the impact of educational games. For example, Zeng et al. [11] used static and evolutionary games on a strategy simulation model to find the causes and solutions to miners’ behavior outside the workplace. Currently, SGs provide benefits such as improved retention, program adoption, and job performance [18,19]. Sipiyaruk et al. [20] recognized that SGs are engaging tools for effective learning because they enhance knowledge and skill. Mokhtar et al. [21] explained that many existing SGs with educational purposes fail to provide mastery content to achieve the game objectives, as they focus more on entertainment. Kara [22] presented the importance of SG in education and its trend in use in recent years. He concludes that various game practices with learning objectives have been used and researched in education. Kristoffersen et al. [23], Mittal et al. [24], and Fleming et al. [25] explored the use of SG for health, education, and other purposes. These studies highlight the importance of continuing research and generating knowledge about SG and gamification.

Emerging technologies, like Virtual Reality (VR), are being applied to SG development for worker education, as demonstrated by Chengmao and Xiaoyu [26], who created a VR system for mining safety training. VR technology facilitates understanding the mining work environment by simulating real situations with high interaction and realism [2]. Game engines enable the creation of realistic VR models, presenting simulation games for training in mine self-escape, exposing the user to danger without being in danger [27]. In addition, VR simulation systems are proposed for training in self-climbing in subway mines, working at heights, and fire scenarios. VR offers significant and sufficient advantages in safety issues [28] and advantages such as immersive and risk-free learning [29].

Building on these advancements, the objective of this article is to review the work carried out on SG development methodologies for the mining sector and their use cases. We begin with the hypothesis that there are surely not many works in this area, so we will extend some of the searches beyond the mining sector, in particular, to the works that have automated these SG methodologies. Thus, the SG developed for training in the mining sector and the methodologies used for its development were reviewed. Then, advances were sought in the automation of these methodologies for the agile development of SGs; however, as no work was found in this aspect for the mining sector, works that have been performed in other areas of applications were analyzed. According to the above, the following three research questions were designed.

Q1. What methodologies have been used to develop SGs for the mining sector? Q1 seeks to review the methods used to develop SGs to train mineworkers.

Q2. What has been achieved in SGs for the mining sector? Q2 seeks to know what SGs have been made specifically for the mining sector since it is one of the most dangerous industrial sectors in the world. There are other high-risk industries such as manufacturing, fishing, electricity, and construction. However, mining has one of the highest mortality rates and the highest number of people exposed to the risk of death [28,29].

Q3. What has been performed in the automation of SG methodologies to generate artifacts? Q3 seeks a general background on automation in the SG design process, which can be adapted to the mining sector, given the lack of jobs of that type in the mining sector [30,31].

There are related works, such as the literature review conducted by Aleem et al. [32], which focuses on the software engineering process life cycle in game development. Checa and Bustillo [33] described how the immersive VR SG research enhances learning and training. Other works presented general approaches to SG design [34]. Some focused on leveraging SG in specific domains, such as early childhood education [35], vaccination [36], and healthy nutrition [37]. Sandí and Bazán [38] reviewed a series of methodological proposals suggested for the design of educational video games.

In SG’s research, literature reviews on learning and training enhancement were found [39]. Regarding methods, processes, and methodologies of SG development, some works about the life cycles of the game development process were found [3,30,34,38]. Other reviews focus on SGs with VR for learning and training enhancement [33]. There are also publications about methodologies for designing SGs [36,40,41,42]. However, their scope differs from ours because they respond to research trends on the life cycle of the SG development process.

Specifically, this review focuses on SG development methodologies, SGs for training, and the automation of SG methodologies, specifically for the mining sector. While there are many studies on SGs, SG methodologies, and the use of SGs in training processes, as noted in the previous paragraphs, no literature reviews on SGs specific to the mining industry were found. Therefore, the main contribution of this work is the development of an SLR on the methodologies used to develop SGs applicable to the mining sector. The methodology used in this SLR has been used successfully in the literature review in different domains; some of them being logistics [43,44], education [17], and agriculture [45]. This systematic review presents SGs that promote training on working at heights, heavy machinery operation, risk situations, mechanized and non-mechanized mine simulations, and mine evacuation systems. This review contributes to the research community’s knowledge about SG training studies and SG development methodologies for the mining sector. One of the main findings concerns the necessity of covering the specification of users’ cognitive functions and emotions across SGs. This finding will allow the development of future research in areas of knowledge such as affective computing, cognitive computing, and behavioral engineering.

This paper is organized as follows: Section 2 presents the methodology for the SLR; this part defines the inclusion and exclusion criteria, sources of information, search strategy, description of the study selection process, and a general analysis of the search. Section 3 presents an analysis of the papers selected for each research question. Finally, Section 4 presents a general discussion about the selected papers, their limitations, and their challenges.

2. Methodology

The methodological process used in this SLR was inspired by the PRISMA methodology [46], which has been successfully used to analyze the literature on the use of computer science in different fields of application (see [43,44] for some examples of the application of this methodology). PRISMA is an accepted standard for presenting evidence in systematic reviews and meta-analyses [47] (see Supplementary Materials for more details on this methodology). The data were collected using the systematic research method, which allowed for an exhaustive bibliographic review.

Our modified methodological approach considered five steps: (1) definition of eligibility criteria, (2) specification of information sources, (3) design of the search strategy, (4) implementation of the selection process, and (5) analysis of the works. The starting point was the specification of inclusion and exclusion criteria, followed by the identification of the databases and search engines to be used. The search terms were defined using the PICOC (Population, Intervention, Comparison, Outcome, and Context) methodology that allows a detailed search, aligned with PRISMA requirements. The use of PICOP is relevant because it helps to clearly structure and delimit the search criteria, motivating other researchers to replicate the review and to determine the scope of the research [48]. The search strings were established for each research question, allowing obtaining relevant publications according to the research questions. We considered two steps to evaluate the papers: the review of each article by title and abstract, and the full-text reading. Finally, a synthesis of the publications that passed the second filter and met the quality criteria defined for the study is provided. Each step of the methodology is detailed below.

2.1. Eligibility Criteria

Table 1. Inclusion criteria.

Criteria	Description
IC-1	Include publications whose titles are related to the development of the SG
IC-2	Include publications in English
IC-3	Include publications from 2010 onwards
IC-4	Include articles in journals or conferences, and book chapters

and

Table 2. Exclusion criteria.

Criteria	Description
EC-1	Exclude publications published on websites
EC-2	Exclude theses
EC-3	Exclude publications whose full text is not available

present the inclusion (IC) and exclusion criteria (EC) of the SLR, respectively. These criteria were chosen to ensure the relevance, timeliness, and quality of the studies included in the SLR. The ICs give priority to peer-reviewed sources such as journal articles, conference papers, and book chapters. The temporal restriction reflects the rapid technological advances in SG development, emphasizing recent studies aligned with current progress in the field. The ECs facilitate the exception of sources that do not meet peer review standards or are difficult to access. In addition, papers whose content cannot be fully verified are excluded, thus ensuring the transparency and methodological rigor of the review.

Table 2 presents the exclusion criteria (EC) for the SLR.

2.2. Information Sources

The data sources include scientific databases such as ACM (Association for Computing Machinery) for its specificity in Computer Science and Technology, SCOPUS as an international database with a high scientific impact, WoS (Web of Science) for its scientific relevance, and Google Scholar as an academic search engine.

2.3. Search Strategy

Two fundamental activities were performed: (1) definition of the search terms and (2) definition of the search string by research question.

2.3.1. Definition of the Search Terms

In this search process, the terms were classified using the PICOP method to determine the SLR objectives.

Table 3. Keywords or groups of terms.

PICOP Method Element	Primary Terms	Secondary Terms
Population	Miner	Mining staff, mining company staff
Intervention	Serious game	Educational game, gamification
Comparison	Not applicable	Not applicable
Outcome	Methodology	Method, models
	Automation artifact	Code generation, development methodology
Context	Mining industry	Mining company, SME mining, mining plant, mining sector

presents each PICOC element with its primary term and the secondary terms that facilitated the definition of the search string for each research question [49]. The comparison element of the PICOP method was omitted because it does not apply to the SLR object of study.

Each primary search term allows for describing a general category of search terms. The description of each general category is presented in

Table 4. General categories of search terms.

ID	General Category	Description
C1	Miner	Terms related to miners
C2	Serious game	Instruments or means that allow the improvement of the training processes in the companies, in this case, serious games
C3	Methodology	Terms related to the methodologies used
C4	Mining industry	Terms of the development field or research sector, which in this case is the mining sector
C5	Automation artifact	Terms related to automated artifacts

.

2.3.2. Definition of the Search String

The search strings were defined for each research question of the study. Each search string is the union and intersection of categories (primary terms) and subcategories (secondary terms) using the logical operators “OR” and “AND”, as shown in

Table 5. Search strings by research question and focus.

Research Question	Description	Search String	Focus
Q1	What methodologies have been used to develop SGs for the mining sector?	(Methodology OR Method OR Model) AND (“Serious Game” OR “Educational Game OR “Gamification”) AND (Miner OR “Mining Staff” OR “Mining Company Staff”)	SG development methodologies
Q2	What has been achieved in serious games for the mining sector?	((Miner OR “Mining staff” OR “Mining Company Staff”) OR (“Mining Industry” OR “Mining Company” OR “SME Mining” OR “Mining Plant” OR “Mining Sector” OR “SMB Company”)) AND (“Serious Game” OR “Educational Game” OR Gamification)	SG in the mining sector
Q3	What has been performed in the automation of SG methodologies to generate artifacts?	((“Automation Artifact” OR “Code Generation”) OR (“Development Methodology”)) AND (“Serious Game” OR “Educational Game” OR Gamification)	Automated SG methodologies

.

2.4. Study Selection

The selection process for scientific publications (including articles, book chapters, and conference papers) is carried out in three phases, inspired by the PRISMA methodology [45]. These phases include the identification of publications in Phase 1, the screening and eligibility of publications in Phase 2, and the selection of relevant publications in Phase 3. The process ensures a thorough and rigorous review of all relevant publications.

• Preliminary publication identification: The search string for each research question is used in the databases and scientific engines selected. Then, the results are filtered by the inclusion criteria corresponding to publications in English and studies after 2010. It defines the list of scientific publications for the next phase.
• Selection by title and abstract reading: In this phase, we examined the documents that met the inclusion criteria and applied the exclusion criteria.
• Selection by reading full text: In this phase, full-text documents were read, and those that met the defined quality criteria were recovered for analysis.

2.4.1. Quality Criteria

In this section, quality criteria (QC) were established for each research question and evaluated by reading the full text of the publications.

• QC-1: Scientific papers and book chapters that present methodologies and detail the development process of an SG, with its respective phases, outcomes, and examples of use.
• QC-2: Scientific papers and book chapters detailing SG developed for the mining sector, their features, usability, development architecture, testing, and stakeholder evaluation.
• QC-3: Scientific papers and book chapters describing methodologies that generate artifacts for developing SGs.
• QC-4: Scientific articles that define SGs how they will be understood in this work, a software that achieves learning objectives and allows the development of specific learning skills [19,21,23].

To apply each criterion to the articles found, the following empirical evidence was identified for each study: For QC1, the empirical evidence consisted of determining whether the publication presented well-defined phases and demonstrated practical applicability in the development of SGs. For QC2, the empirical evidence consisted of establishing whether the publication presented a case study or implementation of SGs linked to the mining sector. For QC3, the empirical evidence consisted of determining whether the publications described tools for the development of SGs. Finally, for QC4, the empirical evidence consisted of establishing whether the publications understood an SG as software that achieves learning objectives, enabling the development of specific learning skills.

2.4.2. Search Conduction

The main search questions of this article (Q1 and Q2) concern the use of SG in the mining sector due to its danger (as mentioned in the introduction). The last question (Q3) complements the first two due to its relevance in software engineering. The search strategy analyzes the progress of each one. Regarding the challenges, they are integrated into future work through SGs to train professionals in the mining sector. Figure 1 shows the publication retrieval process. The search recovered many publications in its initial phase (phase 1). The inclusion and exclusion criteria applied in Phase 1 and Phase 2 allowed the keeping of articles and book chapters whose titles are related to developing SG, and whose keywords match those defined in the search string. Finally, with the application of quality criteria, Phase III gives the final list of publications to be analyzed.

The initial results comprised 1987 studies. In Phase 1, 200 publications respond to each research question in the following manner: 33 publications for Q1, 86 for Q2, and 81 for Q3. In Phase 2, 96 papers were obtained and distributed as follows: 25 focused on Q1, 30 focused on Q2, and 41 focused on Q3. Finally, in Phase 3, 32 publications resulted in the research question: 7 for Q1, 13 for Q2, and 10 for Q3. All papers respond to the quality criteria defined in this phase.

2.5. Preliminary Synthesis

This section presents a description of the publications selected by quality criteria and the preliminary results of the selected studies.

2.5.1. Description of Publications Selected by Quality Criteria

This section presents the 30 selected publications and their relationship with the quality criteria defined in Section 2.4.1. Specifically,

Table 6. Evidence of compliance with quality criteria by publication.

Publications	QC-1 Evidence		QC-2 Evidence	QC-3 Evidence	QC-4 Evidence
	Well-Defined Phases	Practical Applicability	Mining Sector	Artifact Generation	Learning Objectives	Learning Skills
[50]	x	x	x
[51]	x
[52]	x	x
[53]	x	x
[54]		x	x		x
[55]	x	x			x	x
[56]	x		x
[57]			x		x	x
[58]			x		x	x
[59]			x		x	x
[60]			x		x	x
[61]			X		x	x
[62]			x		x	x
[63]			x		x	x
[64]			x		x	x
[65]			x		x	x
[66]			x		x	x
[67]			x		x	x
[68]			x		x	x
[69]			x		x	x
[70]		x		x
[71]		x		x
[72]		x		x
[73]		x		x	x	x
[74]		x		x	x
[75]		x		x	x
[76]		x		x	x	x
[77]		x		x	x
[78]		x		x	x
[79]		x		x	x

describes each publication, considering whether it has the empirical evidence to determine whether it meets the quality criteria. The description is grouped by the research question.

In Section 3, as each study is presented, the quality evidence indicating how well it meets each criterion is also described.

2.5.2. Preliminary Results of the Selected Studies

Preliminary results of the selected studies start with the distribution of papers on SG methodologies for training in the mining sector worldwide. Figure 2 shows that China has the highest scientific contribution in the group of analyzed papers, with eight papers, equivalent to 26.7% of the selected papers. It is followed by the United States (U.S.) with four papers (13.3%).

Figure 3 shows the publication types. Thus, of the selected publications, 16 (53.3%) correspond to journal articles and 14 (46.7%) to conference presentations. In addition, 56.7% were published between 2016 and 2023; the remaining 43.3% were between 2010 and 2015.

The information sources of the publications found correspond to 23.3% (seven articles) for IEEE, 23.3% (seven articles) for Elsevier, and 13.3% (four articles) for Springer. ACM, Taylor & Francis, and MDPI AG represent 20% (6 articles) of the articles found. Advances in Intelligent Systems and Computing, AMexIHC, Bentham Science Publishers, Caltek s.r.l., EDP Sciences, Environmental Engineering and Management Journal, and SAGE represent 20% (six articles) of the studies. Therefore, it has been demonstrated that this SLR provides reliable information since the articles reviewed belong to journals of high scientific reputation, which were peer-reviewed to ensure compliance with quality standards.

3. Findings

The analysis of the works is presented in three sections. Each section responds to the works selected for each research question. In each of these sections, the contributions and research gaps of each article are described. Finally, each section ends with comments and figures summarizing what was achieved for each research question.

3.1. Serious Games Development Methodologies

The following is an analysis of the selected papers that meet the quality criteria established for Q1: What methodologies have been used to develop for the mining sector?

Considering the evolution of coal production, Tang et al. studied the antecedents of complex and dangerous environments in coal production and work accidents inside the mines [50]. Specifically, accidents during production, natural disasters that cause the insecurity of mine workers, and other accidents require working on the issue of safety in coal mines. They presented a methodology for modeling and simulating the complex virtual environment of the coal mine using a three-layer architecture: (1) data, (2) multi-agent, and (3) human–machine interface to create a virtual mining agent and environment. The agent includes reactive (real-time response, perception, and action) and cognitive (symbol and behavior inference) structures. This work presented a guided method for the design of the simulation of the coal mine production behavior. Regarding the practical applicability, the methodology focuses on simulation and training for mining operations and safety, which is a typical case of application of SGs oriented to education and training. Therefore, this publication meets QC-1 and QC-2. However, it lacks an application example that allows for validating the stages of the proposed methodology.

Grabowski et al. [51] presented a method for immersive VR systems with results applied to 21 coal mine workers exposed to two simulations: moderate and high immersion. They defined phases and practical applicability relevant to the development of interactive simulations for training purposes, which can be extrapolated to SGs in similar contexts. Consequently, the fulfillment of QC-1 and QC-2 is demonstrated. The results of this work encouraged the owners of training centers cooperating with Polish mines to introduce VR in the basic training of younger miners. As for developing simulation environments, this work lacks a life cycle with processes, activities, and deliverables typical of a simulation mode software development method.

Aslan et al. [52] developed GAMED (diGital educAtional gaMe dEvelopment methoDology), as a methodological proposal for developing games based on the life cycle of DEG. GAMED consists of four phases (see

Table 7. Summary methodologies.

Paper	Methodology	Phases	Validation	Applicability	Technological Tool
[50]	Method for design simulation of the coal mine production behavior	VR system definition Agent modeling Multi-agent system construction Planning Implementation Validation	Specific case of the process in which a virtual mining agent evades a transportation vehicle	Simulation and training for mining operations and safety	Interactive simulation with virtual agents for task planning, movement control, and real-time production behavior
[51]	Method for immersive VR systems	Training scenario design Implementation of VR Evaluation Statistical analysis Feedback integration	Testing with experienced miners to gain insight into training experience	VR-based training for subway miners	Specific virtual training scenario related to blasting work in subway mines
[52]	GAMED	Game design Game software design Game implementation Game-based learning and feedback	Not specified	Development of DEG projects of all sizes	Not specified
[53]	Behavlets method	Game analysis Identifying characteristics and behaviors Coding of Behavlets Feature selection and structuring	No application case, but an example of use with a Pac-Man type game with data from 100 players	Workshop evaluation with expert designers using Gears of War	Not specified
[54]	VR method developed for the generic CMBF plant	No phases	Specific use case of the VR model applied to a generic CMBF plant in Brazil	Improvement of training and emergency simulation processes in industrial and educational environments	VR tool for the green iron industry, specifically for generic CMBF plants
[55]	iPlus	Identification Pedagogical objectives Participatory design Refinement Evaluation and validation	Case study of SG Educaplay	iPlus has a wide and flexible applicability for the design of SGs for educational purposes	SG through a user-centered design approach [55] SG for Labor Inclusion of People with Intellectual Disabilities [80]
[56]	3D method + CAD simulations	3D scanning Computational simulations Collision zone identification Subway infrastructure redesign Visualization	Example applied in Polish mining, focusing on auxiliary transport by rail and monorails	Practical application in the mining industry	Not specified

) and twelve processes. Thus, GAMED has well-defined phases and has practical applicability for the development of educational SGs. Thus, this publication meets QC-1. GAMED guarantees quality at every stage of the game’s life cycle. Thanks to the step-by-step guide for developers and stakeholders, products are finished within the set time. Unfortunately, this work lacks a case study that would allow the research community to apply it to other fields of study.

Cowley et al. [53] developed an approach called Behavlets, which is a methodology with well-defined phases and practical applicability to the development of SGs. The methodology consists of a systematic process that integrates expert knowledge and psychological theories to derive game characteristics that describe player behaviors in an interpretable and theoretically grounded manner. Hence, this publication meets QC-1. Particularly, this work combines expert knowledge in behavioral issues and game design patterns to recognize and model players’ behavior based on various metrics such as deaths, in-game choices, avatars, player characteristics, and play times. They developed an example using a Pac-Man-style game with data from 100 players to analyze behaviors through Behavlet coding, identifying traits, and shaping feature sets based on psychological theory and player modeling. This methodological work does not present an application case that validates the process of developing SGs (for more details, see Table 7).

Vieira et al. [54] created a VR model of a coal-fired mini-furnace plant (CMBF). They presented a methodology to build the VR model that includes well-defined and concrete phases that can be applied to the development of SGs. Its practical applicability is for the creation of immersive educational environments and interactive simulations. Therefore, this publication complies with QC-1 and QC-2. The VR prototype was built using data from various software and hardware, including detailed engineering designs of auxiliary units and main equipment. The VR model allowed researchers to explore all parts of the CMBF plant and have technical discussions with users, improving information management, communication, safety, maintenance procedures, and engineering project development. This VR platform has become a professional education tool for universities, museums, and industries, benefiting students, workers, and users. This model presents a prototype that lacks specifications for the development phases of the VR system.

Carrión et al. [55] presented a methodology for developing SGs focused on user participation, called iPlus. The contribution of this publication is based on a participatory focus, in which experts and users involved in developing the SG concept are identified. This methodology presents phases and activities that guide the development process of an SG (see Table 7). It integrates learning objectives and participatory game design to enhance users’ cognitive skills. Although it has not yet been applied in the mining sector, it shows strong potential for training-oriented SGs. Thus, the publication meets quality criteria QC-1 and QC-4. Although its participatory focus is one of its main features, at the time of implementation, it becomes a limitation since each phase requires expert users for its development. Nevertheless, this makes it useful for development teams with sufficient resources or large projects.

Rozmus et al. [56] developed a method that combines 3D laser scanning technology with computer simulations performed in a computer-aided design (CAD) program to solve transportation problems in subway coal mines. They presented a methodology focused on the redesign of selected areas of subway transportation routes in coal mining by integrating 3D scanning, computer simulations, and VR. This methodology comprises well-defined phases that ensure its practical applicability in SG development. Although this work does not explicitly describe the development of recreational or educational applications, it shows its applicability for the creation of SGs, especially in areas such as operator training, industrial safety, or process simulators. They tested the method in the operating conditions of a real crossing in the Premogovnik Velenje mine (Slovenia). Therefore, this publication meets QC-1 and QC-2.

Table 8. Summary of methodologies and technologies used in publications selected for Q1.

Paper	SG Methodology	Not Specified	Multi-Agent Technology	VR Technology
[50]	Method for design simulation of the coal mine production behavior		x	x
[51]	Method for immersive VR systems			x
[52]	GAMED	x
[53]	Behavlets method	x
[54]	CMBF			x
[55]	iPlus	x
[56]	3D method + CAD simulations			x

summarizes the results of the seven papers responding to research question Q1. Table 8 shows four publications that explain models and methodologies applied to the mining sector using VR technologies. In addition, one publication is focused on developing models applying AI, specifically multi-agent technologies. Finally, three publications do not specify the technology used to model SGs.

VR technology enables safe and realistic training environments for mineworkers, requiring specific methodologies for SG design. Key contributions include VR simulations for surface mining, multi-agent systems, and dynamic environments to enhance workers’ reactions and skills without real-world risk. Publications applying user evaluation methods and models to measure the effectiveness of learning in SGs with specific approaches, such as scenario design, story design, character design, and gamification, were also found. Some problem domains solved by these publications were (1) simulation of mining production systems, (2) real-time monitoring of mining workers’ performance, (3) accident reconstruction using CAD technologies, (4) training of mining workers with the use of VR, and (5) natural catastrophes incurring mining workers’ insecurity. Table 7 shows that three methodologies do not specify the use of a particular technology. However, according to these results, there is the possibility of trying other technologies (for example, Augmented Reality and Mixed Reality).

Finally, we present a summary of the methodologies, their phases, examples of use or validation, and the technological tools developed (see Table 8). In this table, we see the diversity of phases that each methodology has (as well as their purposes), as well as the diversity of uses, tools, and validation strategies that have been made.

3.2. Serious Games in the Mining Sector

The following is an analysis of the selected papers that meet the quality criteria established for Q2: What has been achieved in SGs for the mining sector?

Stothard and Hengel [57] developed a mining initiation training module as a pilot study aimed at improving mining initiation in the South Australian hard coal rock mining industry, through the development of an SG prototype. Specifically, they presented a simulation module based on an SG designed to promote the development of practical skills related to mining safety in the area of working at heights, a reason that demonstrates the compliance of the publication with QC-2 and QC-4. This module comprises five sub-modules: Correctly Erected Scaffolding, Incomplete Scaffolding, Open Excavation, Light Globe Change using Ladders, and Elevated Work Platform. The pilot module was first tested at the Adelaide City Training Center with 42 trainees. The objective was to expose the module to a real audience and verify its usability from the trainer’s and trainee’s points of view. The trainees answered real-time questions about the SG simulator in this simulation environment. Of the 42 trainees, 20 were exposed to the SG simulator, and the other 22 were exposed to PowerPoint slide training. The results with the simulator were positive since it caught the attention and was easy to use. However, the knowledge results do not demonstrate the ease of learning.

Assuming the challenges of a decade ago by SCG users, today, more realistic simulators are taking advantage of VR. Thus, Andersen et al. [58] developed and tested a METS VR, including a more realistic scenario for coal mining employees, with direct applicability in the mining sector. It is specifically designed to train subway mine workers in emergency evacuation procedures without interrupting normal mine operations, which represents a significant benefit to this industry. For this reason, this publication complies with QC-2 and QC-4. This immersive experience allows better-quality training and low cost for companies. Moreover, thanks to its portability, it was used for training in evacuation and emergency, and the mine did not require suspending production activities. The simulator uses VR technology to create an immersive and realistic environment for miners undergoing evacuation training. The article describes the design and implementation of the simulator, as well as the results of the simulator tests and evaluations. However, in the implementation phase of the simulation, the development methodology applied has not been identified.

Li et al. [59] developed a system under browser/client architecture and applied AI techniques to the game. They presented the VR platform for mine safety training, providing a tool for safe coal mine work and emergency training. The platform incorporates non-playable characters (NPCs) with artificial intelligence (AI) that guide the user through subway tours and accident experiences, such as gas explosions, roof falls, and fires, to build users’ cognition and learning skills. This publication shows compliance with QC-2 and QC-4. This application allows players to connect to the server and get the 3D streaming from the server in the cloud. The server manages the events and status of the virtual world and sends the VR rendering streaming to the client. It was successfully applied in the Experiment Laboratory for undergraduates at China University of Mining and Technology—Beijing. The results of this paper provided a reliable, safe, and low-cost tool for safety and on-the-job training in coal mines, avoiding high-risk subway environments. This work lacks a methodology in the process of implementing the tool.

Zujovic et al. [60] focused the research on the quality of heavy equipment operator (HEO) training as a key to improving safety and reducing injuries related to equipment operation in the mining industry. They presented a VR SG oriented to safety training on risks associated with rock falls in subway mining. This game offers a training experience that raises awareness of safety issues and promotes hazard identification without exposing the user to real risks, thus demonstrating its practical utility in mining contexts. In terms of learning objectives and skills development, the game strengthens safety awareness and promotes competencies related to risk detection and mitigation in mining environments. This publication shows compliance with QC-2 and QC-4. This project used a 360-degree camera, the open-source WordPress 6.7 platform, and Unity3D 6.0 software to create materials and tools for HEOs’ safety training to help trainees better understand the pre-shift machine safety inspection. The computer-based safety task training website developed in this research has been tested and implemented at a surface coal mine in the Southern U.S. Three main phases were used to build the proposed HEO training: (1) collection of the materials; (2) construction of the computer-based task training (CBTT), and (3) construction of the VR application, in which interactive training courses were worked on, capturing 360-degree content (images and videos) and collecting training materials to incorporate into the computer-based training, and VR application. In the development of the CBTT application, a flowchart describing the various components of the CBTT is evident. However, this work lacks a requirements analysis process for constructing the VR application.

Oliveira et al. [61] developed a virtual training system based on their research on VR systems for locomotive engine maintenance training. They implemented a VR simulator for industrial engine maintenance training, specifically implemented in a Brazilian mining company. The simulator achieves learning objectives such as parts identification and engine disassembly procedures. In addition, the simulator promotes the development of motor and cognitive skills related to industrial maintenance. For this reason, this publication complies with QC-2 and QC-4. This system has some functionalities, such as an initial game presentation view, user input, a part selection environment for disassembling, a VR game screen, and a scoring scheme. The tool is a tutor-coordinated and easy-to-use first-person game; even a person with little experience in video games can interact with the VR system. The solution presented is effective and low-cost. This virtual trainer was evaluated with 59 users, resulting in satisfactory usability of the system, immersion in the game, and user comfort and experience. This paper shows the development engine used to develop the application and the modeling technologies employed in constructing the simulator. However, this paper lacks a formal flow of activities that would allow developers to follow a guided process for application in another field of study.

Huang and Hou [62] explored the shortcomings of methods and systems in coal mining. They also presented an architecture of the multi-service monitoring platform for mine safety, the hardware design, and the detailed definition of related modules. This platform combines video, voice, and data using the GENE-8310 industrial control board to improve communication and supervision in mines while ensuring safe production. For this reason, this publication complies with QC-2 and QC-4. Then, through the latent form simulation test in the laboratory, it is demonstrated that the platform can realize the design goal that all services under the mine are IP-based. The system is based on a three-layer architecture: (1) mine environmental parameter acquisition layer, (2) mine site monitoring layer, and (3) site management layer. This paper presents an architectural model for constructing a multi-service monitoring platform but lacks elements to describe the platform’s training focus.

Zhang [63] presented the construction of an intuitive VR training system based on a Head-Mounted Display (HMD) for training drilling scenarios in underground coal mines, composed of five parts: user, I/O devices, VR engine, software, and database, and finally, the tasks, which for this case is the drilling of wells based on operating procedures. They used VR to train in situations that are complex to practice in the real world, such as mine rescue or disaster evacuation, demonstrating that VR-based SGs have real and practical applicability in the mining sector. This VR system is used to train tasks such as hole drilling, facilitating users to immersively learn by doing and develop skills needed to operate in real mining scenarios. For this reason, this publication demonstrates compliance with QC-2 and QC-4. This paper discusses the features and elements of VR technology that can ensure authenticity in user training. However, it does not address the formalized software development process and analysis of pedagogical objectives, which are crucial for the effective implementation of VR training.

Concerning developing systems with VR technology to improve the skills and survival of miners in the face of coal and gas explosions, Tan et al. [64] designed and developed a VR system at the China University of Mining and Technology. They presented a virtual training system for the mining sector, developed using Virtools, which focuses on safety training related to coal and gas release accidents. The system aims to improve survival and evacuation skills during landslide scenarios by reinforcing practical knowledge of self-rescue techniques, rational escape planning, and the correct use of personal rescue equipment. For this reason, this publication demonstrates compliance with QC-2 and QC-4. The system implemented 3D visualization of the process of coal and gas outburst and underground escape, creating a sense of immersion in the user’s reality, with escape routes being impressive, understanding the method of self-help and the different types of explosions that can occur in a subway mine, understanding the accident process and reducing the consequences of a mining accident. This VR system makes it possible to make coal mine safety training interactive and interesting. Furthermore, it compensates for the deficient conventional safety schemes in the mining sector. The virtual survival training system has three modules: the accident simulation module, the evacuation training module, and the evacuation evaluation subsystem. This paper presents the development process of the VR system. However, it lacks a formalized method.

Following the line of work of VR systems, Xiaoqiang et al. [65] developed a fully mechanized system for a mining context with VR technology. They presented a VR system applied to mining engineering for the simulation of a fully mechanized mine face. Although not directly termed as an “SG”, the system has clear applicability in miner training, student practices, and rescue simulations. Concerning learning and skills development objectives, the system simulates mine face production processes (cutting, haulage, support, etc.) and provides a real-time immersive environment that facilitates understanding of work team coordination. For this reason, this publication demonstrates compliance with QC-2 and QC-4. The system has a geometric modeling process and element partitioning techniques, which allow for improving the virtual view of the coal mine. The system has roaming functions, interaction scenes, equipment operation, and production process simulation, establishing a training approach for miners in case of a catastrophe.

Ivina et al. [66] developed a system for the dynamic construction of a virtual mine model for underground coal mining operations. They implemented the software module “VR Mine” as a serious VR-based training system that includes several components such as a database, integration with the “3D Mine” module, tools for editing 3D models, and a training script module. This system is designed to simulate unpredictable emergencies that users must solve, promoting safe and controlled training in the mining sector. Its real applicability is in mining, and it is not a fun game, but a game for training purposes. For this reason, this publication demonstrates compliance with QC-2 and QC-4. The system allows logical interaction with equipment, the definition of preventive emergency response plans, and the definition of emergency response plan measures. Bednarz et al. [67] demonstrated through experiments how immersive VR provides an interactive space for training purposes in underground coal mining operations, in surface mining, with techniques applicable to monitoring mining environments with sensor networks. They develop an immersive VR (IVR) application with a serious and educational purpose, designed specifically for the mining industry and with the potential to be used as an e-learning and training tool. The application promotes experiential learning, as the user can physically manipulate 3D objects and observe behaviors. This publication shows compliance with QC-2 and QC-4. The above papers lack a development process guided by a methodology to understand the inputs and outputs in each phase of implementing a VR system with training scopes.

Cai et al. [68] developed an emotional behavior model for virtual miners based on cognitive appraisal theory and an interactive virtual environment to simulate the behavior of miners in underground coal mines. They defined a serious virtual environment that simulates human behavior in underground coal mines, aiming to improve worker safety and training. While the publication does not explicitly state specific learning objectives or targeted skills, the environment recreates accident scenarios and models safety-related behaviors to enhance training effectiveness, support decision-making, and increase miners’ awareness of potential risks. This publication shows compliance with QC-2 and QC-4. To achieve the development of the proposed model, they used the typical cases of coal mine safety accidents in the past 20 years in China. They generate virtual environments for the proposed models to generate more believable behaviors based on personalized emotional states. This proposed model automatically assesses the emotional significance of perceived events related to the targets. In addition, it derives a plausible emotional state and alters their behavior accordingly to create a more believable virtual coal mine environment. This model can be used as a training tool to analyze the resulting risk factors for subway mining workers. Experimental results show that the proposed models can create more realistic virtual coal mine environments in real time to simulate human behavior, resulting in subway accidents. The simulation’s construction lacks a mechanism that defines the user’s cognitive and emotional components.

Liu et al. [69] designed an autonomous mining and transportation system. They developed MetaMining, a platform that uses the Metaverse concept to integrate and enhance physical mining with an advanced virtual system. MetaMining enables the simulation of mining processes, optimization of operations, improvement of safety and efficiency, and facilitates the coordination of autonomous mining vehicles and machinery. This platform indirectly supports technical training and operational skills development in a safe and advanced environment. Therefore, this publication shows compliance with QC-2 and QC-4. They also presented the different ways in which computer-generated content can be presented, and the different technologies that allow virtual and physical interaction. This paper aims to demonstrate that the proposed meta-mining architecture can achieve high-efficiency and high-security mining in the physical world through the interaction between the physical world and the virtual world. They propose an architecture composed of the human world, the physical mining system, and the virtual mining system. Thus, the architecture contains (1) digital twins, (2) digital natives, and (3) surreality. In this way, the architecture allows users in the human world to exchange information with the virtual and physical mining systems and manage them.

When analyzing the results of each publication, we highlight that most of the studies analyzed in group Q2 lack a formal methodology. This observation is supported by Table 6, which shows the publications’ compliance with the quality criteria. Specifically, we see that all articles in group Q2 do not meet QC-1, but we also note that these articles did not use any methodology (or at least did not indicate it) for the implementation of SGs. Therefore, the incorporation of good methodological practices for the development of SGs in the mining sector is recommended.

The summary of the works for the research question Q2 is as follows: The technologies used in SG for training in the mining sector are mainly VR, with 69.23% of the total number of publications analyzed, 3D simulation with two articles (representing 15.38%), and AI technologies, Internet of Things (IoT), and cloud rendering with the remaining as shown in

Table 9. Summary of technologies in selected papers for Q2.

Papers	VR	AI	3D Simulation	IoT	Cloud Rendering
[57]	x
[58]	x
[59]		x
[60]			x
[61]	x
[62]				x	x
[63]	x
[64]	x
[65]	x
[66]			x
[67]	x
[68]	x
[69]	x

.

The development of SGs for training in the mining sector using 3D simulation, VR, and AI considers topics such as (1) safety in mining work, (2) training of miners, (3) underground mines, and (4) coal mines. Nine papers (69.23%) related to topics 2 and 4, which are developments of SGs for training workers in coal mines. Topics 1 and 3, with four papers for each topic, show the use of SGs to work on occupational safety issues, specifically for underground mines, as shown in

Table 10. Summary of mining topics in SGs.

Papers	Mining Work Safety	Miner Training	Underground Mines	Coal Mine
[57]	x	x		x
[58]	x	x	x	x
[59]	x	x		x
[60]		x		x
[61]		x		x
[62]	x		x
[63]	x	x	x
[64]		x	x	x
[65]	x	x		x
[66]		x	x
[67]		x	x	x
[68]	x	x	x	x
[69]	x

.

Figure 4 shows that Unity 3D is the most used game development engine for VR implementation. Unity is used in demonstrations with virtual mine models, virtual tours, and immersive VR experiments in subway and surface mining. They represent 46.2% of the publications selected for research question Q2. The remaining percentage was distributed in papers related to the use of Unreal Engine in the construction of training systems and virtual mining models, DI-Guy for modeling human behavior in a multi-agent virtual environment, and Virtools together with Quest 3D in interactive visualization and 3D simulation of accidents, training systems, and evacuation assessment.

3.3. Automated Methodologies for Artifact Generation

The following is an analysis of the selected papers that meet the quality criteria established for Q3: What has been performed to automate methodologies for generating artifacts?

Borror and Rapos [70] presented MOLEGA, a framework to solve the code generation and implementation process with JavaScript, delivering a functional artifact based on the game model, which is focused on the creation of web-based educational card games, intended to be used in educational environments to support learning. It has a practical applicability since it allows users without advanced technical knowledge to create customized educational card games used directly in educational environments. This operative applicability is based on the fact that the tool facilitates the generation of games accessible from any browser-based device and that the content of the games can be easily adapted for different educational purposes. In addition, MOLEGA allows the automatic generation of software artifacts from visual models created in its guided environment. The evaluation of MOLEGA consisted of a plan to verify positive and negative errors, validating that the generated code is correct for its use. MOLEGA is more than a model editor; it is a framework that ensures that the models produced are valid and consistent, presenting a constant check of the game’s rules to be developed. The methodology is automatic, checking that if a user does not comply with any requirement of the metamodel, the panel shows a red X instead of a green one and informs the user how to solve the problem in detail. If the model the user intends to generate contains errors, the generation function is disabled so that no valid code is generated. The work demonstrates the feasibility of using DSMLs to create web-based educational games with minimal technical knowledge; its only limitation is that code generated in JavaScript has disadvantages in interoperability with other technologies.

Hamiye et al. [71] proposed a framework for generating automated artifacts (functional code) from UML and DSL models for SGs and defining the SG development process for assessment (SGA), delivering to the pedagogue a general game structure using low-level technical language such as domain-specific language (DSL). They demonstrate through a proof of concept how two different games can be designed and developed using the modeling framework. The framework structure starts with the idea of the game to be developed, combining visual and textual language to design an SG. The project used UML to define the game structure and a domain-specific language for the game logic with scoring rules. The code generation was based on each model designed with tools and code generators and combined with extra logic.

Chevaillier et al. [72] presented the framework called MASCARET (Multi-Agent System for Computer-Aided Real-time Environments and Training), a tool for developing intelligent virtual environments based on UML metamodeling and with operational semantics to the models, which can be executed in several VR platforms, meeting the requirements of content and system-oriented semantic modeling. MASCARET is included in this review because it presents practical applicability in the development of interactive virtual environments. Some practical applications may include the simulation of human activities in complex environments and the design of virtual laboratories for education. Using MASCARET, two virtual environments with their respective intelligent virtual agents were developed. The article demonstrates the potential of the MASCARET platform to create sophisticated virtual environments with realistic agent behavior and interaction. However, it lacks examples that evidence the code generated from the metamodels. The work completed by Kotani et al. [73] is an extension of SuperSQL to generate VR scenes from 3D object information stored in a relational database. This extension allows users with little knowledge of VR tools to create 3D scenes of a virtual museum using only a simple SQL-like query. The system has practical applicability in the automatic generation of 3D virtual environments from relational databases using SuperSQL. This system allows the reduction of workload and complexity for users, facilitating the creation of virtual spaces with different layouts and layouts, with utility in contexts such as virtual museums, exhibitions, and online commerce environments, as well as educational applications. It also supports the generation of C# and XML code artifacts for SGs and educational applications in VR, facilitating their integration into different platforms and projects. To generate the virtual museum with 3D scenes with extended SuperSQL queries, the user must first collect the 3D objects to be displayed and the metadata of the objects to be stored in the relational database. Then, the virtual scene is generated with the information inserted into the database. Once the query is executed, the saved data are recovered and generated in two XML files describing the data and its hierarchy. Finally, C# files are executed in Unity, and XML files are processed to display the scene. A limitation of this work is that the resulting products are only for museums.

Bucchiarone et al. [74] developed a gamification design framework, applying engineering modeling techniques. They proposed a structured process with clearly defined phases, supported by a modular and multilevel architecture, which facilitates the development of artifacts for SGs. A tool is defined to design gamified applications based on three layers of modeling: (1) definition of gamification elements, (2) specification of the composition of the elements necessary for the game design, and (3) game instance (implementation and execution). This project produced code for games using GaML 7.06 and game instance modeling language (GiML 3.0). This work limited the code generation to JAVA 23, with the instance classes defined through the game instance concept. The tool has practical applicability in the development of SGs that allows the generation of artifacts for their construction. It is created to support the design and implementation of gamified systems using a model-driven modeling approach. In addition, it facilitates domain experts to design games in terms specific to their area, without focusing on implementation details, thus favoring automatic code generation and reuse.

Following the approaches based on models that allow growth in virtual environments, Céspedes et al. [75] developed a methodology that generates natural user interfaces for natural interaction in humans. This methodology seeks to automate communication activities, such as dialogues, requests, complaints, and suggestions. They include the delivery of information and the development of voice interfaces, establishing rules to support a software tool that generates code for multiplatform voice interfaces. The methodology abstracts common voice interface components and supports automatic code generation for multiple platforms, including VoiceXML 3.0, XHTML + Voice 1.0, and Kinect v2. Its practical applicability is focused on the development of dialogue systems, automated services, and other cross-platform voice applications. To establish the methodology, they worked with three tools, VoiceXML, XHTML + Voice, and Microsoft Kinect, with the Microsoft voice synthesizer v11. In this work, the limitation is that the platform uses a specific modeling language based on the model-driven paradigm (MDA). Choi et al. [76] used the inter-operation of virtual worlds and simulators and proposed an automated methodology for developing SGs with three basic processes. The processes are game loop analysis, game agent design and development, and parameter tuning. They used a high-level architecture to guarantee the inter-operation between a simulator and a virtual component. This approach allows the generation of automation artifacts for SG development to achieve effective interoperability and support the developer’s work. Then, they generate an SG from an existing base component, reducing efforts and providing more real experiences. They showed a practical applicability in extending and improving existing SGs for military training by facilitating integration with constructive simulators. A fundamental characteristic of this methodological proposal is that it makes it easier for a developer to synthesize the federation when the game agent and a simulator are available. This paper lacks the model specification with the necessary features to extend a particular SG and a case study to validate the code generation.

In integrating the game modeling language (GaML) with the Unity game engine, Matallaoui et al. [77] proposed the development of SGs based on models. They defined a model-driven approach based on GaML and Unity as a work with practical applicability in SG development, especially in the automatic generation of code for achievement systems within such games. The modeling language is only descriptive and does not generate executable software artifacts, solving this with a complete architecture based on models that allow designing and generating building blocks for SG, with a validation system from an existing SG. This approach facilitates communication between designers and developers to reduce the effort and time needed to implement achievement systems since the generated code does not require manual modifications. With the use of GaML, portability is guaranteed; models are independent of the system architecture since abstract concepts are described without any relation to the system or the architecture where they will operate. An important feature is that developing the game’s gamification concept does not require expert designers to operate the application. The validation of the approach was through a real case study (a serious application “Stop Smoking”) demonstrating the practicality and effectiveness of the approach in a concrete scenario. Its limitation is that the design of SGs is only given from the UML language, being of general use, so it loses the advantages of domain-specific languages, which should be used in developing SGs based on models.

De Lope et al. [78] developed a set of metamodels for designing educational video games using UML language, with adaptations made between students, teachers, and the technical design team. The proposed methodology allows the development of artifacts hierarchically, starting with high-level act diagrams and moving toward lower-level diagrams such as scene, action, and dialogue. The methodology also emphasizes integrating educational content into the plot and narrative of the game to facilitate assimilation and retention. Uranos is a game example that demonstrates the methodology’s use and validation. The methodology proposes a seven-phase approach to developing educational video games, starting with game decomposition and ending with dialogue design. This work does not generate the complete code of a video game from UML diagrams, so in the future, it is necessary to study tools that facilitate this code generation process, which positively impacts the development time of educational games. This work describes a practical applicability in the design of SGs, especially in narrative and adventure-based games. They also considered automatic code generation for the development of these games. Additionally, the work’s validation included the modeling of a real educational game called Uranus that served as a case study to show the feasibility and value of the work, including educators in the design and facilitating collaboration between multidisciplinary teams.

Solís et al. [79] defined a language for video game modeling and code generation using business process modeling notation (BPMN) on multiple platforms. Specifically, they presented VGPM to model the “game loop” or the main logic of the game through a graphical language based on BPMN with specific elements for video games, facilitating the creation of complex logic without the need for programming. Therefore, it is evident that VGPM has practical applicability in SG development and the automatic generation of code for games. This language has the characteristics of game logic and reduces video game development costs. The modeling language is specifically for the game loop logic, with specific elements used as some constraints and validations specific to the video game business domain. The processes generated by the users are based on BPMN standards, allowing them to operate with other BPMN tools. This work developed a generation engine that converts the elements in the business model into source code and, combined with the GADE4ALL editor, allows the generation of fully functional video games with a game logic graphically configured by the user. This paper does not compare its approach with other video game development approaches. A comparison could help establish the relative strengths and weaknesses of this approach.

In research question Q3, methodologies, frameworks, and extensions were found. The methodology term is related to a set of steps that guide a process to achieve a product, in this case, the development of SG. An extension corresponds to functionality added to a methodology to improve a given process. A framework corresponds to implementation processes, techniques, or stable models. Figure 5 presents the crossover with the artifacts generated from the publications found in Q3. Five publications (50%) were related to the term methodology. The other 50% were extensions and frameworks for code generation, gamification design, SG definition using SGA, and the development of intelligent virtual environments. Four publications (40%) of the total selected for Q3 studied component generation, three publications (30%) studied the use and production of UML diagrams, and three publications (30%) generated code for SGs.

After analyzing the results of the publications selected for Q3, we observed that most articles are not related to the mining sector. The publications reviewed in this section do not specifically focus on tools for high-risk contexts such as mining. However, several of the tools presented could be adapted to mining, as they are applicable in high-risk contexts (see Table 6). They facilitate the development of SGs and could be adapted to mining with appropriate design considerations. In particular, we have added

Table 11. Tools with high adaptation potential for the mining sector.

Paper	Tools
[70]	DSML-based tools Tools to create simulators and SGs
[71]	Multiplayer and Collaborative Evaluation Support Tool
[72]	MASCARET Framework VRX-OCL for spatial restrictions Autonomous Agent System with Action Logging
[73]	Unity game engine Extended SuperSQL system for automatic generation of 3D environments
[74]	GDF based on Model-Driven Engineering (MDE) and using JetBrains MPS
[75]	VoiceXML standard Multimodal technology such as Kinect for flexible and safe interaction
[76]	Virtual application as In-World Editor with specialized construction simulators, interoperating via High-Level Architecture
[77]	Unity game engine GaML Model-Driven Software Development and Xtext
[78]	SGs diseñados con metodología UML
[79]	Specific BPMN-based languages adapted to mining

with the list of possible tools that can be adapted from each study to the mining sector. In general, code automation is the most easily adaptable and offers the opportunity to reduce SG development costs by training personnel in high-risk environments such as mining.

The first conclusion from this section is that some frameworks generate code from UML metamodeling, BPMN, game-based models, SGA, and engineering modeling techniques. The methodologies provide the generation of software components or their reuse to guide an SG generation process. The second conclusion is that the implementation of extensions is low, generating a space for research and development of functionalities in defined and stable methodologies.

4. Discussion

4.1. Review Summary

In this section, a Sankey flowchart [70] is used, with the objective of exploring the trend of the reviewed articles on SGs for training in the mining sector. Particularly, this flow allows us to analyze the articles studied from different dimensions (groups) of interest for our work, defined in the upper part of Figure 6. Thus, specifically, Figure 6 shows the 30 articles reviewed in four groups (dimensions); the first group corresponds to technologies relevant to our study. The second group characterizes the techniques and models most used in the articles. The third group corresponds to the objective of each research question. Finally, the last group is related to what specific topic in the mining sector is considered in each article. When an article was not found to have a suitable type in a dimension, it was not classified.

The first group, which shows the technologies most frequently used in the 30 analyzed studies, is explained based on the categories illustrated in Figure 6. VR has long been applied in education [71] and has achieved high adaptability [72]. Querrec et al. [81] used modeling systems to design virtual learning environments for vocational training. Kamińska et al. [82] brought together the most recent applications of VR related to education, engineering, and health care. Peng et al. [83] combined traditional school education with VR, recognizing VR as an immersive device to isolate the individual from the real world. Gómez and Trefftz [84] conceived VR from the human–computer interaction perspective to support learning processes. This perspective motivated them to design an immersive experience with low hardware costs for developing countries.

VR technology with operator training simulators is widely used in the mining industry to increase workers’ awareness and skills and help improve their qualifications. Pedram et al. [85], Pedram et al. [86], and Pedram et al. [87] showed VR technology as one of the most widely used for training in various operations and risky events. Bergamo et al. [88] demonstrated that VR provides effective mining safety training as a promising method of improving training.

Most of the publications in this SLR addressed VR technology, and 4 of the 16 papers on VR technologies correspond to SG development methodologies. Tang et al. [50], Grabowski and Jankowski [51], and Vieira et al. [54] developed methodologies for miner training with VR. They showed the complexity of exploring complete knowledge of the level of safety supervision of the entire mining sector. Rozmus et al. [56] worked to build methodologies to assess the mining behavior of miners in underground mines and simulate computationally underground works.

Eight publications responded to the development of SGs for the mining sector, of which five focused on training mine workers through immersive simulation. Stothard and Hengel [57] focused on training on working at heights [64]. Moreover, Andersen et al. [58] designed escape systems and simulators for evacuation training. Zhang [63] and Xiaoqiang et al. [65] studied immersive and interactive training systems. Two studies have been developed on human behavioral risk to simulate situations that generate real emotions in the player in immersive virtual environments [65,66]. In addition, Oliveira et al. [61] were devoted to developing SGs to train operators for industrial engine maintenance.

Four studies applying VR respond to Automated Methodologies (AMs). Chevaillier et al. [72] used semantic modeling and knowledge representation to generate components in virtual environments. Choi et al. [76] proposed the interoperation of simulators and existing VR applications, in which they developed a case study for nuclear/biochemical evacuation training. Ref. [73] developed a system for generating VR scenarios based on 3D objects. The object data were stored in relational databases, allowing declarative queries to extract information from each object in the museum. Céspedes et al. [75] worked on tools used in developing virtual environments (e.g., VoiceXML, XHTML + Voice, and Microsoft Kinect) for modeling vocal UI based on the knowledge of the context. Finally, one study applied AI for SG development [59], which uses a VR system with a mining catastrophe animation based on Cloud Rendering technologies.

Regarding three-dimensional (3D) simulation systems, the relevant contributions of authors who focused on providing solutions for the mining sector by developing 3D visualization systems and 3D simulation systems are presented. Tan et al. [64] developed an accident simulation module using 3D Max and Virtools for modeling and animation. They focused the module on the 3D visualization of the coal and gas blowout process. Ivina et al. [66] focused on 3D visualization in underground mining, realized for the dynamic construction of underground coal mining operations (e.g., the interior of a mine, the mining activity, and the installed technical equipment). They used more than 30 3D objects to complete the development of the visualization of operations inside a mine. Zujovic et al. [60] proposed an interactive system of training in the operation of HEO to reduce operator injuries. They studied mining safety issues and presented an HEO training tool, which was tested and implemented at a surface mine in the U.S. They used 3D software (WordPress 6.5 and Unity3D 5) to develop the project to simulate mining operations environments.

The most relevant work on web technologies found in the review was MOLEGA [70], which used Domain Specific Language (DSML) for players to create models representing educational card games. The MOLEGA web editor allows a fully guided model creation process and supports the import of metamodels based on the Instructional Modeling Language (IML). Ref. [62] contributed to monitoring to provide mining safety by using IoT technology and integrating audio, video, voice, and data. They used the TCP/IP protocol to provide a secure network for information transmission in subway mines. They postulated the possibility that the services of an underground mine could be networked and centralized in a single system. Hamiye et al. [71] presented a methodological framework for modeling and generating EMS code, with UML meta-modeling for game structure and a DSL for logic (interaction schemes and points). This approach needs additional code to fit the generated code with any game engine to achieve a fully functional game.

4.2. Challenges

Based on the analysis of the selected papers, the following challenges were divided into thematic groups (SG methodologies, automation, and mining sector). Furthermore, within each thematic group, they were divided into subcategories to highlight those that we consider to have greater or lesser impact: (a) High Impact, (b) Medium Impact, and (c) Low Impact.

4.2.1. SG Development Methodologies

These challenges revolve around Q1, on the methodologies defined to develop the SGs, and are categorized by impact as follows:

(a)

High impact

• Represent hazardous environments to stimulate safety awareness: The studies presented simulations of accidents and risk situations using VR technology in countries such as India, the U.S., China, and Australia [48,50,64]. However, no environments in Latin America or other regions have been developed to simulate risk situations and stimulate safety awareness [51].
• Establish mechanisms that involve the player and encourage him/her to remain in SG practice: It is necessary to maintain the user’s attention and concentration in meeting the learning objectives established in the SG. The studied SG development methodologies presented phases of generating the game idea or concept, team and role conformation, pedagogical objectives, game development, improvement, and feedback [89]. However, in the review and analysis of the literature, no publications were found that integrate mechanisms that involve the player and stimulate him/her to remain in the SG practice.
• Define mechanisms for specification of cognitive capabilities with SG development: The SG methodology analysis did not delve into the user’s cognitive capabilities. Most methodologies are focused on the work team (designers and developers). The studied SGs focused on using virtual simulation systems, machinery operation, coal mine monitoring, and accident reconstruction. Therefore, it is necessary to propose a mechanism to streamline the development of SGs that promote cognitive skills in a specific domain [90,91].

(b)

Medium impact

• Define adaptive mechanisms for SGs: Classically, the SG methodologies assume games that cannot be adapted to the user and simply work based on what is specified in their design [92]. For this challenge, it is possible to take the examples described by Streicher et al. [93], which explains how adaptivity can be implemented in specific application scenarios, such as motion-based games or personalized learning games. There is also a whole theory of emerging SGs that could be explored in this area [42], in particular, allowing game dynamics to emerge from interactions with the players themselves.
• Define methodologies for developing SGs for effective guided training: In our review, we did not find methodologies that specify the necessary aspects for guided training, nor mechanisms to evaluate the effectiveness of the training processes. Some progress was found with AI for specifying guided behavioral characteristics of game characters (specifically, non-playable characters) with conditions and rules programmed into the AI to mimic the character and copy behaviors in a game character [94].
• Characterize factors for user empathy: Most SG development methodologies focus on the game lifecycle [30,38,50] and specific game genres [78,89]. However, it is necessary to develop SG methodologies that guarantee to keep the player’s attention, motivate the player to reinforce the knowledge for its complete understanding, and ensure that the player voluntarily repeats the training sessions. For this challenge, it is necessary to review the study by Bachen et al. [90], who proposed exploring the interrelationships between factors that may influence empathy and interest in learning more about a specific subject is pertinent.

(c)

Low impact

• Specify intelligent mechanisms to analyze/evaluate user behaviors: This challenge aims to include in the methodology the design of a mechanism that allows profiling/evaluating user behavior. This challenge is closely linked to the previous one, but here, the conception of an intelligent tool that determines and analyzes user behavior is explicitly considered [44,80]. For this challenge, it is possible to use the result of Syufagi et al. [95], whose focus is the modeling of SG systems and classification of motivational behavior with Petri nets.
• Develop strategies for evaluating player performance: The review showed poor data collection from SG users. Knowing the players’ data and the variables present in an SG, such as performance, movements, problem-solving methods, decision capacity, and response time, is a promising challenge. Some approaches proposed approximate solutions, but nothing on evaluating user performance in specific activities in an SG [96].

4.2.2. Automated Methodologies

The challenges presented in this section focus on generating artifacts derived from automated methods, frameworks, and methodologies that streamline the SG development process. Some artifacts can be codes, UML diagrams, and software components to advance the SG development process, which are categorized as follows:

(a)

High Impact

• Implement modeling systems to improve productivity in SG development: Software development is expensive, time-consuming, and labor-intensive. The field of SG development requires automated tasks to streamline and deliver initial prototypes in less time. Some studies have proposed automation up to the game design document [44,97]. There is a need to improve productivity in SG development, presenting specific challenges such as project resource planning, requirements definition, and tracking the learning curve of programming languages and technologies for SG development.

(b)

Medium impact

• Develop functionality to add behaviors to playable characters: The characters in a video game play a role and seek to empathize with players by generating highly immersive environments. Developing immersive environments involves improving the quality of graphics and storage space, considering the increase in production by generating characters in large volumes [80]. Regarding the quality of the graphics, it is necessary to make the models more realistic, which implies more storage space. Therefore, it is necessary to reuse mechanisms that lighten the character design process with high-quality rendering.

(c)

Low impact

• Generate VR scenarios for SG development: Immersive virtual environments require 3D modeling software, robust game engines (Unity and Unreal), image editing software, programming environments (C#, C++, JavaScript, among others), and hardware (e.g., Oculus Quest and Oculus Rift). SG development using VR technology is costly due to the tools required to design real scenarios. Some scenarios represent the elements of the game environment (e.g., environment, sound, characters, story, and gameplay) from DSM models. The models of [70,72] generate challenges such as the possibility of creating and integrating code for stable SG engines.

4.2.3. Mining Sector

According to the topics explored in this SLR, the development of SGs in the mining sector focuses on training mine workers. Therefore, there are challenges and opportunities such as:

(a)

High impact

• Promote safety training awareness in the mining sector by developing cognitive enhancement mechanisms. These safety training mechanisms are necessary to train miners with little experience in mining work [94].
• Develop mechanisms to improve the cognitive capabilities of mine workers to perform their work responsibly and reduce the risk of danger, using the methodological process applied in similar projects [98].
• Define training to develop the ability to perceive strange odors, preventing the death of mining workers. The developed countries have solutions like the one suggested by [70] with a costly and complex architecture to implement in underdeveloped or poorer countries.

(b)

Medium impact

• Develop SGs for evacuation and rescue systems due to gas escapes based on the models for developed countries [99,100,101] but less expensive to implement.
• Specify mechanisms that allow the mine to be supervised without risk, avoiding entry into the interior of the mine. For that, it is necessary to generate safe working environments for the miner, where the miner can receive indications in case of emergency from outside the mine.

(c)

Low impact

• Study and propose other architectures adapted to underdeveloped or poorer countries’ economic and geographical conditions.
• Design of multi-agent systems aimed at mining safety. This challenge can be inspired by results from previous work, such as the work of Wimolsakcharoen et al. [102], in which a computer role-playing game was proposed to support the exchange of perceptions and knowledge between stakeholders based on multi-agent systems. Also, it can be inspired by other previous works related to the use of SGs for the management of natural resources in the fields of agriculture, biodiversity, water, and forests [103,104,105]. Finally, works based on multi-agent systems for rescue in emergencies (for example, in earthquakes or natural disasters), such as the one presented by Takahashi and Shimizu [106], are also initial sources of inspiration for this challenge, to adapt them to the mining sector.

Overall, these challenges serve as a guide for SG developers and mining stakeholders, with very concrete practical implications, as follows:

• For simulation game (SG) developers, the challenges allow identifying methodological improvements, future tool developments, possible adaptive extensions, specific modeling tools, and specific strategies for optimizing development processes, among other things. Thus, it is possible to define a technological development plan based on these challenges.
• For mining managers, the challenges allow establishing an SG implementation plan that promotes safety awareness, and cognitive and affective training can help reduce accident rates and improve decision-making in high-risk operations.
• Finally, policymakers can benefit from this knowledge to establish public policies for the mining and technology sectors that support the development and financing of safety training technologies tailored to the economic and social reality of each country.

5. Conclusions

This is the first SRL focused on the intersection of SGs, mining, and methodological automation to our knowledge. This study examined 30 articles that met the eligibility criteria to answer the research questions. SG development methodologies were reviewed, and the importance of having a complete process that guides SG developers in each phase of the game life cycle was concluded. Also, the need for a specific methodology to describe users’ cognitive functions to be developed through SGs was concluded (none was found to meet this specification).

In the search for answers to the first and second research questions, the results show that SG developers have been dedicated to building simulation systems and, to a lesser extent, to designing multi-agent systems aimed at user safety. However, this subject has not been worked on for SGs in the mining sector. As the physical safety of the mining worker continues to be a challenge for the mining sector, this is one way to analyze it. Furthermore, it highlights the importance of developing SGs to evaluate user performance and behavior in situations that generate emotions and behavioral reactions, which also represents a challenge for the mining sector.

In the search for answers to the third research question, the results highlight the importance of reusing components for code generation. Also, advances in metamodeling (graphical representation) were found to automate artifact generation methodologies for SG development processes in the mining sector. Other metamodeling notations used by researchers in the reviewed studies were MDA, BPMN, MDE, and DSM. Finally, there was evidence of the need to extend the automation of graphical models and develop plugins to reduce code incompatibility with SG development engines.

Finally, considering that mining training seeks to achieve a learning objective, the potential use of SGs in education and other scientific fields of mining is recognized and noted. This SLR allowed the identification of challenges or research opportunities in areas such as pedagogy (learning objectives and learning outcomes), psychology (cognitive and executive functions), and affective computing (AI, emotion analysis).

The most notable findings of this SLR are the following. There is a significant use of SGs in the educational field that impacts training processes in the mining industry. Likewise, there are a significant number of methodologies with complete development phases, some based on the life cycle of educational games, but with little use in the mining field. Finally, this review provides a compendium of highly interesting works. The challenges identified can guide the mining research community and managers on how to introduce SGs as training mechanisms in high-risk industries such as mining.

This review has limitations concerning works that analyze users’ emotional training. The documents studied only used cognitive appraisal theories and fuzzy logic techniques to simulate the response behaviors of mine workers. Another limitation in the review is that no methodologies were found that integrate aspects of the specification of cognitive functions with the affective part of the user. It did not allow us to analyze these aspects, which are very relevant due to the importance of emotions and emotional self-regulation in digital learning environments such as those provided by an SG. Another area for improvement is that it is necessary to analyze the methodological components that can be reused to generate automatic code. The few in the literature were compatible with the most used SG development engines (Unity and Unreal Engine).

Finally, the following research lines in SGs are recommended: first, expand methodological frameworks for developing SGs, incorporating the user’s cognitive and emotional components; second, learning environments that utilize SGs should be created to validate user behavior in occupational risk situations; third, automate the development process of SGs using MDE to enhance efficiency and lower costs for companies and game development teams; fourth, unsupervised training approaches with a feedback mechanism to self-adjust SGs should be implemented; and fifth, design patterns based on player behavior for game design processes should be identified to improve games’ effectiveness. Also, in the mining context, we recommend that future SG developments focus on the construction of simulation systems and multi-agent systems aimed at user safety training. We also recommend advancing the automation of artifact generation methodologies for SG development processes based on metamodeling (e.g., in graphical representations).