Enhancing the Decision-Making Process through Industry 4.0 Technologies

Abstract

In order to meet the increasingly complex expectations of customers, many companies must increase efficiency and agility. In this sense, Industry 4.0 technologies offer significant opportunities for improving both operational and decision-making processes. These developments make it possible to consider an increase in the level of operational systems and teams’ autonomy. However, the potential for strengthening the decision-making process by means of these new technologies remains unclear in the current literature. To fill this gap, a Delphi study using the Régnier Abacus technique was conducted with a representative panel of 24 experts. The novelty of this study was to identify and characterize the potential for enhancing the overall decision-making process with the main Industry 4.0 groups of technologies. Our results show that cloud computing appears as a backbone to enhance the entire decision-making process. However, certain technologies, such as IoT and simulation, have a strong potential for only specific steps within the decision-making process. This research also provides a first vision of the manager’s perspectives, expectations, and risks associated with implementing new modes of decision-making and cyber-autonomy supported by Industry 4.0 technologies.

1. Introduction

Socio-economic developments are transforming the way work is done and decisions are made within companies. These developments naturally impact the performance of companies that have become more agile to comply with new market requirements. To this end, many companies are seizing the opportunities offered by emerging technologies, especially those related to Industry 4.0 [1], and many works have studied the relationships between Industry 4.0, lean or continuous improvement and their impact on social systems or, more generally, on sustainability issues [2,3,4,5,6].

A German government program to increase the competitiveness of its manufacturing industry is behind Industry 4.0 [7], which was announced at the Hannover Fair in 2011 [8]. Since then, Industry 4.0 concepts have been pushed by various governments under different names and more than 100 definitions have been proposed [9]. Although different views and distinctions coexist as to which technologies are included in Industry 4.0, the Boston Consulting Group [10] identifies nine main pillars, namely, autonomous robots, simulation, horizontal and vertical integration of information systems, the Internet of Things, cyber security, the cloud, additive manufacturing, augmented reality, and big data and analytics. Furthermore, cyber-physical systems (CPS) are presented as one of the most significant directions in the development of computer science and information and communication technologies by many authors [11,12,13]. CPSs integrate other 4.0 technologies and interact with humans and other autonomous subsystems at all production levels through many new modalities [11]. CPSs contribute to enhancing the collaborative and cognitive aspects associated with the different stages of the decision-making process [12].

Companies have often started implementing these technologies to increase their profitability and productivity but sometimes neglect the human dimension. If the latter is impacted by the deployment of these new technologies, it turns out that human factors also impact these deployment processes in turn. This is also the case when introducing new technologies to support decision-making processes. While benefits may accrue [1], they seem to be maximized only if multiple organizational adjustments are integrated, especially those related to dynamic abilities and human factors [14]. In practice, these adjustments are rarely studied or considered all together.

To identify and understand the adjustments to be made to increase the benefits of Industry 4.0 technologies, it seems important to study the relationship between technologies and their use in decision-making processes. In particular, the question arises to what extent these technologies strengthen the employee’s empowerment and facilitate interactions between employees and managers. These considerations have become more and more prominent so that Industry 4.0 has already given way to Industry 5.0 [15]. Industry 4.0 is considered to be technology-driven whereas the Industry 5.0 concept, announced by the European Commission, is value-driven, with three interconnected core values: human-centricity, sustainability, and resilience [16]. In this context, a safe and inclusive work environment must be created to, among other things, prioritize autonomy, which is considered one of the fundamental rights of workers [16,17,18,19]. It is to be noted that numerous thinkers or observers see the main characteristic of Industry 5.0 as bringing the human touch back to the center of decision making through the collaboration between humans and machines. In this context, it seems that issues of empowerment and autonomy in decision making carried out by humans assisted by new technologies will become particularly crucial in the future [20].

From an autonomy perspective, Industry 4.0 technologies are tasked with enabling industrial machines and systems to become context-adaptive and autonomous [21]. At the same time, employees appear to gain autonomy at work by using such technology. Some works based on the concept of human cyber-physical systems (H-CPS) attempt to identify human–automation symbiosis work systems. For example, Romero et al. [22] offer a typology of operators 4.0 based on how the physical, sensory, and cognitive capacities are reinforced by Industry 4.0 technologies. However, this work does not specify how these enhanced capacities modify autonomy at work and improve decision making in an operational context. More generally, the advances provided by Industry 4.0 create significant opportunities for better decisions to be made [23], particularly on the shop floor with effective data-driven decisions [24,25]. However, these potential benefits brought about by Industry 4.0 raise new research questions that remain to be clarified:

• What is the potential of all the new technologies associated with Industry 4.0 to strengthen the decision-making process?
• What are managers’ expectations regarding the enhancement of the different parts of the decision-making process with Industry 4.0 technologies?
• Will the enhancement of the decision-making process by Industry 4.0 technologies impact the evolution of the autonomy of operational teams and systems?
• How will the answers to these questions evolve as the level of integration of Industry 4.0 principles within companies increases?

This article aims to answer the first research question mentioned above as a priority. The current literature does not specify the potential of Industry 4.0 technologies to enhance the decision-making process. Therefore, our research objective is to identify the contribution of all Industry 4.0 technology groups to enhance the decision-making process in an operational context.

To achieve this, we will present a literature review of Industry 4.0 technologies and decision-making models in an operational context. We demonstrate the need to conduct exploratory research on the relationships between all the technological groups associated with Industry 4.0 and the different steps of the decision-making process. Therefore, to pursue our research objective, we carried out a prospective study by consulting experts using the Delphi method. We have also applied the Régnier abacus [26], an original technique that consists of collecting, using a colored panel, the opinion of experts based on statements expressed in precise, concise, and relevant terms. Finally, we will present the results of this study with insights on the potential evolution of the autonomy of teams and operational systems induced by the introduction of these new technologies.

The paper is structured as follows. First, Section 2 presents a review of the literature on decision-making models and research analyzing how new technologies can improve them. We will demonstrate the need for further research to clarify the relationships between all Industry 4.0 technology groups and all decision-making process steps. Section 3 describes our research methodology, which coupled the Delphi method with the Régnier abacus technique. Section 4 is devoted to the presentation and discussion of the results. Section 5 presents the research limitations and prospects, and Section 6 concludes with the main results obtained.

2. Literature Review

The CEFRIO group [27] argued that Industry 4.0 brings together a set of tools promoting the improvement of processes, products, and services through decentralized decisions based on real-time data acquisition. According to the National Institute of Standards and Technology (NIST) [28], this smart manufacturing environment raises key questions, including the reorganization of work in the physical and virtual enterprise, the modes of regulation between the different stakeholders, and the evolution of current decision-making processes. While there is no single definition of the concept of Industry 4.0, improving the decision-making process appears to be a recurring focus and a primary objective in the deployment of new technologies [29,30,31].

Industry 4.0 brings out real-time decision making in a decentralized way [32] but coordinated as a global system to bridge together men and machines [33]. These developments promote the flexibility and temporality of decision making at the operational level by increasing responsiveness and autonomy [27]. Romero et al. [22] proposed a typology of operators 4.0 and distinguished several types of operators assisted by Industry 4.0 technologies whose enhancement induces an evolution of responsibility in conducting operational activities and decision making [34,35]. However, this work does not explain the specific ways in which the different Industry 4.0 technologies enhance operators’ decision making.

Highlighting the difficulty for manufacturing companies to establish a strategy for deploying Industry 4.0 technologies, Osterrieder et al. [36] proposed a smart factory model structured around eight distinct thematic perspectives, one of which concerned the decision-making process. They also noted that the challenges associated with decision making are common to many of these perspectives. The authors pointed out the need for further research on the decision-making process and to collect new evidence of the usefulness of decision making supported by data in manufacturing and reinforced by Industry 4.0 technologies.

Human decision making has been studied in many fields, including psychology and management sciences. This process has been described and analyzed through extensive research in various operational contexts [37], strategic [38] and crisis situations [39,40]. The intuitive and analytical strategies involved in these decision-making processes have also been studied through laboratory experiments or field observations to shed light on judgments and phases of decision making under complex conditions [41,42,43]. Simon [44] was among the first to propose a decision-making model, which is the most concise but also the most comprehensive characterization of a rational approach to decision making [45]. This model describes decision making in three phases: Investigation, Design, and Selection. Using the ideas of Simon’s model and analyzing 25 decisions from different companies, Mintzberg [38] proposed a model for strategic decision making in companies. Its decision-making process is defined as a set of actions and dynamic factors that begins with identifying a stimulus to action and ends with a specific commitment to action. The three phases proposed by Simon are described in terms of seven central steps called “routines”, supported by three sets of routines, decision control, communication, and politics. This model is non-sequential and offers the ability to bypass certain steps and interrupt the process or provide feedback.

The so-called “naturalistic decision-making” perspective (NDM) [37,46] was born from the desire to describe the actual decision-making process. All the results of the studies conducted on the NDM emerged in the 1980s. The authors focused their research on the biases and limitations of human decision making, particularly in situations of time constraints [47] or crisis [46,48]. The results demonstrate the need to move from “normative” models describing how rational decisions should be made to models describing the decisions made in reality [49]. Some work highlighted the peculiarities of decision making in naturalistic contexts [50] as well as the unrealistic nature of assumptions that underlie the “rational choice theory” often used in the explanation of decision making [46]. In an operational context, individuals are regularly subject to constraints that reduce their time to process information or to perform complex assessments. This limits the number of available choices identified and biases the evaluation of different options.

If these studies approach the decision-making process pragmatically, they remain solely focused on human decisions without considering any technological support [51]. Other studies describe decision-making models where decision-making activities are carried out by humans or machines (through automation) but do not necessarily lead to optimal decision making [52,53]. However, these works do not link or recognize the different technologies that can assist or strengthen the process [54,55]. Other authors have proposed idealistic decision-making models, particularly in work related to the development of artificial intelligence and intelligent agents, including BDI (belief–desire–intention) models [56,57]. These models, inspired by models of human decision making, are involved in designing artificial decision-making systems. These models are based on specific technologies, including simulation techniques, big data analysis, and artificial intelligence [58].

The DMN (Decision Model and Notation) standard was developed by the Object Management Group (OMG) [59]. It aims to bridge business process models and business logic models by introducing a decision requirements diagram that defines the decisions to be made in business processes, their interrelationships, and their business logic requirements. The associated model can be used to model human decision making, automated decision-making requirements, or implement automated decision making. Group decisions are always better than individual decisions [60]. In this regard, DMN models can describe collaborative organizational decisions, their governance, and the required business knowledge. This standard is dedicated to operational decisions made as part of day-to-day business processes rather than strategic decisions with fewer rules and representations. Hasic et al. [61] pointed out that DMN has only been studied and implemented statically, despite the dynamic nature of modern knowledge-intensive systems. Models of changing decision patterns have not received much attention so far. Therefore, this type of model does not accommodate a changing and uncertain environment for which decision rules, input data, and business knowledge cannot be established in advance. Therefore, these models do not apply to any type of decision and are unsuitable for operational decisions leading to specific solutions in response to unknown or poorly controlled situations [61]. In addition, some articles attempt to make a connection with research on decision support systems (DSS) [62] or demonstrate how certain technologies facilitate the implementation of this standard [63]. Nevertheless, none of them encompasses the possibilities offered by Industry 4.0 technologies.

In addition, many authors highlight the potential offered by one or more technologies to enhance particular steps of the decision-making process. The analysis of big data and artificial intelligence facilitates the recognition of problems by strengthening sensory capabilities through machine learning methods, allowing the recognition of images, speech, text, or the detection of unusual situations through the analysis of massive data flows [64]. Certain failures can be recognized by CPS equipped with appropriate sensors [65,66]. The adoption of artificial intelligence requires the processing of massive data whose capture and storage is facilitated by IoT and cloud computing. Taking this step enhances the diagnosis of certain situations or the search for solutions by allowing the discovery of hidden patterns and unknown correlations [67]. Stojanovic et al. [68] proposed a new concept of self-aware digital twins combining big data analysis and simulation to monitor the operation of a system continuously, understand its current behavior, detect opportunities for improvement, and simulate a hypothetical process (“what if” analysis). This allows the determination of consequences if the problem is not solved correctly for an extended period of time.

Other authors have pointed out the potential of technologies to strengthen a larger part of the decision-making process associated with well-targeted implementation conditions. For example, Simon et al. [69] developed a technique based on data collected on an agri-food production chain by IoT systems to determine the optimal maintenance procedure among a set of possible alternatives. This expert evaluation system is based on a multi-criteria decision model created using the fuzzy analytical hierarchy method. However, this specific solution only enhances certain key steps in a specific decision-making process. Krueger et al. [70] describe the STAMINA robot system in which Industry 4.0 technologies help to increase robots’ autonomy for performing kitting tasks. These robots are integrated into the Manufacturing Execution System (MES) and operate in a shared workspace between humans and robots to handle abnormal situations spotted by sensors and a vision system. Several sources of errors have been identified and can be handled automatically by the robot. If this complex system enhances the decision-making process, thus conferring greater autonomy on the robot, it remains specific to a given task. In addition, these autonomous robots cannot handle situations that are unknown or problems for which a solution is not known.

The models that deal with the decision-making process are partly based on Industry 4.0 technologies, such as big data [71], or focus mainly on peculiar activities or support for operations, such as maintenance [25], or are mainly interested in strategic decisions [23]. The question of the impact of Industry 4.0 on the decision-making process has so far focused only on the disparate tasks associated with decision making, not on the process as a whole [30]. To date, none of these models considers the opportunities offered by the joint contribution of different Industry 4.0 technologies to the whole decision-making process [72].

In conclusion, research on the decision-making process is mostly focused on or inspired by human decision making. On the other hand, the literature associated with Industry 4.0 offers numerous articles describing examples or proposals for enhancing the decision-making process by a given technology or groups of technologies. These are, however:

• Limited to the enhancement of only partial steps of the decision-making process;
• Concern specific and non-generalizable decision-making processes; and/or
• Do not consider the possible contribution of all technology groups to strengthening the different steps or parts of the decision-making process.

To fill this gap, the objective of this study is to identify the contribution of all Industry 4.0 technology groups to the enhancement of the decision-making process in an operational context by conducting a Delphi study using the Régnier abacus methodology, as explained in the following section.

3. Research Method

This study aims in the first place to study the potential of Industry 4.0 technologies to enhance the decision-making process of operational teams and systems use. Given the current limitations of the literature on this point, the study is qualified as exploratory. For this, the combination of the Delphi method and the technique of Régnier’s abacus appears as a relevant approach.

3.1. The Delphi Method

The Delphi method is recognized as a structured method for obtaining and organizing the opinions of a group of experts from a decision-making perspective, exploring a complex subject, or developing models [73]. The Delphi method is widely appreciated for its ability to advance empirical knowledge and group judgments that lead to the emergence of consensus or dissensus on a subject. It is defined as “an iterative process used to collect and distill expert judgments using a series of questionnaires” [74]. Delphi implementation typically involves two processes: the process of recruiting experts and the multi-time communication process of data collection, called “iterations” [75]. As specified by Rowe and Wright [75], each iteration aims to refine the data collected in the previous iteration and involves a controlled return of the responses. The participants are informed of the other participants’ answers while their anonymity is preserved. To ensure the experts’ participation and minimize the risk of bias, anonymity is essential here. Indeed, it is essential to compare the opinions of the various experts on the subject to achieve the objectives of the research. Still, due to multiple interests, direct communication between them is not possible due to the need to ensure an open reflection which requires an exchange free from any factors other than the sharing of distinct points of view and the creation of knowledge. This iterative process stops when the researcher considers that he has answered the research question [74]. In our case, we managed to obtain stabilized responses after two rounds. These steps ensure the technical validity of linking the experts with each other to stabilize the judgments.

As mentioned by Skulmoski et al. [74], the Delphi method is flexible. Thus, through the literature, many adaptations in its operationalization have been made, particularly with regard to the number of iterations carried out and the data collection methods used. Given the complexity of the research topic and the need to promote more comprehensive data collection on a prospective topic [76], the exploratory data collection method was chosen. For this, we combined the Delphi method with Régnier’s abacus.

3.2. The Régnier Abacus

Régnier’s abacus is an original expert consultation technique that uses a color panel to intuitively collect experts’ opinions about precise, concise, and pragmatic statements. As a business intelligence tool, the advantages of this technique are the speed with which opinions can be summarized and the colorful visualization of results that facilitate decision making [26]. To express their opinion, experts must choose from seven colors:

• Green: the expert completely agrees with the statement;
• Light green: the expert agrees with the statement;
• Orange: the expert’s opinion is mixed;
• Light red: the expert does not agree with the statement;
• Rouge: the expert does not agree with the statement at all;
• White: the expert cannot answer; and
• Black: the expert does not want to answer.

The three main colors (green, yellow, red) indicate transparency in the answers, while white and black indicate opacity. The information collected appears in the form of a colored diagram that brings a complementary dimension to written or oral language. This view makes it possible to quickly obtain clear information. Therefore, the advantage of this technique is the speed of the synthesis of opinions and a colorful visualization that facilitates decision making. This study used the open-source Color Insight (http://colorinsight.fr/ accessed on 22 December 2021) solution to create the questionnaires, collect the answers, and organize a colorful representation.

The combination of the Delphi method and the Abacus of Régnier will be referred to as “Delphi–Régnier” in the following sections of this document. This combination is often used to quickly collect and summarize expert opinions [9,77], to facilitate debate and decision making [77,78], and foster creativity [26,79]. We will describe the conditions for implementing this study in the following subsections.

3.3. Selection of Experts

The Delphi method is characterized by the consultation of a group of experts, a commonly appointed “panel of experts” where the “expert” is defined as an “actor with recognized skills in a field and responsible for contributing to the elaboration of a judgment” [80]. Since the results of a Delphi study are essentially based on the opinion of the persons consulted, particular attention must be paid to the constitution of this panel [81]. Experts must be selected according to three criteria:

• Their experience;
• Their familiarity with the object of study; and
• Their level of knowledge of the characteristics of the object.

To include both the perspective of the actors interacting with the operational teams in the composition of the panel of experts, we have chosen professionals using technologies (called industrials), creators of specialized digital solutions (called integrators), and, finally, academics. To be considered an eligible participant for this study, the person had to meet four criteria:

• Register in at least one of the different categories of participants;
• Have held this role(s) or function(s) or have a minimum of three years of experience in an Industry 4.0 related position;
• Have held that role(s) or function(s) in the three years preceding the study period; and
• Have held this role(s) or function(s) in a private company or in a public institution (for academics).

The second criterion ensured a minimum level of experience, deemed necessary to accurately represent the category associated with the participant, while the third criterion ensured that this experience was recent enough for it to be relevant at the time of the study. Finally, our main concern was to obtain as complete a set of perspectives possible, as well as diversity and balance in the groups represented on the panel of experts.

Depending on the studies, the size of the panel can vary. It is not imposed a priori but will depend on targeted areas and objectives [82]. Mitchell [83] recommends a panel of at least ten experts in studies using the Delphi method. The work of Ashton [84] has shown that the size of the group of experts for a consultation study is around 11. Some other studies have shown that between five and 11 experts ensures sufficient reliability, and that beyond 13 experts per group the average error of the group hardly decreases [79]. On the other hand, it can be noted that most of the recent studies combining the Delphi method and Régnier’s abacus are based on a panel of about twenty experts [77,78,85,86]. Considering all the positions held during the three years preceding the time of the study, the panel of experts was finally composed of 24 experts distributed as follows: eight industrials, eight integrators, and eight academics.

It should be noted that the “industrial” experts selected in the panel are all directly involved in Industry 4.0 deployment projects within their company or a group of companies as decision makers having been managers or practitioners themselves. In this context, they have been led to identify the needs of managers and practitioners and make the links with the objectives and developments identified as strategic within their organizations. The “integrator” experts are also regularly asked to collect the expectations of managers and practitioners to ensure that the solutions they sell and deploy are correctly received and used by the teams using them.

3.4. Survey Creation

The construction of the initial questionnaire submitted to the experts in the first round is a key step in the study, which significantly guides the areas in which the Delphi–Régnier study will generate ideas [9].

A first version of the questionnaire was structured around the enhancement by Industry 4.0 technologies of the decision-making 4.0 model in an operational context (see Figure 1) proposed by Rosin et al. [87].

The ten technology groups (see Figure 2) proposed by Danjou et al. [27] were selected to classify Industry 4.0 technologies. Indeed, the authors have taken up and enriched the classification of Rüßmann et al. [10], already very widely cited. In addition, similar to our approach [72,87], they proposed an Industry 4.0 technology deployment model that builds on the levels of capacity (monitoring, control, optimization, and autonomy) formulated by Porter et al. [88].

This led to the identification of 20 items structured around two prospecting axes:

• The possible contributions of the 10 Industry 4.0 technology groups to strengthening the steps of the decision-making process; and
• Managers’ expectations in terms of enhancement of the decision-making process by Industry 4.0 technologies.

This first version of the question was submitted to a first test panel of industrials and academics attached to the professional group “Operational Excellence and Supply Chain” of Arts et Métiers alumni, which brings together students and alumni of the Arts et Métiers engineering school, the largest European network of alumni of a major engineering school, and a group of industrial alumni of the Ecole des Arts et Métiers and academics working in the fields of operational excellence and supply chains. The group aims to enable observation of the evolution of the prospects and opportunities in this professional branch and help transfer knowledge in this field.

This first test brought out the need to link the evolution of the autonomy of teams and systems at the operational level with the deployment of Industry 4.0. This point was particularly expected by the industrialists solicited within this test group to provide a means of better understanding the issues associated with the reinforcement of the decision-making process by Industry 4.0 technologies; the latter insisted in particular on the need to make a link with the logic of collaboration and empowerment. To ensure the proper participation of the industrial experts who were subsequently asked to respond to the study, items related to the evolution of the autonomy of teams and systems at the operational level were added.

To meet these requirements, the study’s steering committee decided to adopt the model of autonomy at work proposed by Bourdu et al. [89]. This was obtained at the end of a work analyzing autonomy in “emerging” work organizations (including Lean Management, Liberated Enterprise, and Responsible Enterprise) based on a think tank bringing together academics and industrials. This model of autonomy at work revolves around three dimensions:

• Dimension 1 is task-oriented, in which context autonomy involves the latitude for teams and operational systems to be able to define their own tasks, the sequencing of their tasks, methods of execution, the pace of work, and tools to be used;
• Dimension 2 defines the power for operational teams to exert an influence on the organizational environment by participating in the improvement of work organization and by influencing decisions concerning their work or the modes of cooperation necessary for the proper performance of the work; and
• Dimension 3, finally, measures the involvement of operational teams in the governance of their company through social dialogue or negotiation (in relation to simple information and consultation), the degree of influence on the sharing of the value created, the implementation of a participative management mode, and the presence of employee representatives in governance bodies.

These three dimensions delimit a space of empowerment, direct participation, capacity for influence, and making decisions at work within which the autonomy entrusted to teams and operational systems is built. To integrate all the dimensions, additional items have been structured around:

• The link between enhancing the decision-making process and the evolution of the autonomy of teams and systems of operations; and
• Managers’ expectations vis-à-vis the capability of Industry 4.0 technology to increase the autonomy of operational teams in the tasks they perform, in organizations and in governance.

Based on this work, the Steering Committee selected ten questions/statements, bringing the list to a total of 30 questions/statements (presented in Appendix A), also referred to as “items” in this document. All the statements have been carefully examined iteratively by the authors to reduce the risk of misinterpretation.

3.5. Iteration Structure

The experts answered the initial questionnaire of 30 statements using the Color Insight platform. They had to give their opinion with the Régnier’s abacus. Participants were also asked to justify their replies with a short comment. To enrich the study, the experts also had the opportunity to submit additional statements.

From the expert votes on each claim submitted, an item matrix was generated, as shown in Appendix C. This matrix of items helped classify and visualize the statements from the most feasible to the most unfavorable. Dissensus items appear in the middle of the matrix. The matrix was generated according to the “classic mode” proposed by Color Insight: the color weights are 5 for dark green, 4 for light green, 3 for orange, 2 for light red, and 1 for dark red. As there is no standard threshold in the literature, it was decided that consensus was reached when 60% of the responses in favor (green) or against (red) were observed.

At the end of the first iteration, the steering committee reviewed and synthesized the votes and comments. The second iteration could then be launched with the same participants, based on the following documents:

• A summary document presenting the results of the first round and allowing the experts to compare their answers;
• A detailed report presenting the distribution of votes and the anonymous areas for each item in order to minimize the risk of distortion commonly described in the literature [90]; and
• A refined questionnaire of 26 items, provided in Appendix B.

The main trends of opinion among the experts could be confirmed at the end of this second iteration and the steering committee decided to end the study. It should be noted that 21 of the 24 experts in the first round responded to the questionnaire in the second round. Although this decline in participation is one of the weaknesses of the research method [90], the representativeness of the panel was maintained around the following distribution: seven industrials, seven integrators, and seven academics. The results thus obtained at the end of this second iteration remain consistent with those resulting from the first iteration.

4. Findings and Discussion

Following the Delphi–Régnier study, the steering committee synthesized the experts’ answers. During the first round of the study, the questions were prepared by the steering committee, but for the next rounds it was the experts who proposed a list of statements. After analysis of all the data, we identified four themes to group the opinions:

• Autonomy 4.0;
• Decision-making process 4.0: managers’ expectations;
• Decision-making process 4.0 and the level of integration of Industry 4.0 principles; and
• Enhancement of the decision-making process through Industry 4.0 technologies.

In the next subsection, we present a synthesis of the experts’ answers according to the four themes identified. We used the following formula to reference the elements presented to the experts:

(RX-IY): reference to questionnaire X and item Y.

For example, reference R1-I2 refers to questionnaire number 1 and item number 2, “Operational teams and systems will need to be more autonomous to meet future challenges”. The full questionnaires are available in Appendix A and Appendix B and the item matrixes generated from the expert votes are shown in Appendix C and Appendix D.

4.1. Autonomy 4.0

The answers to items R1-I1 and R1-I2 allow us to estimate the gap between the current level of autonomy entrusted to operational teams and systems and the expected or desirable level induced by the deployment of Industry 4.0 technologies. The experts indicate that managerial practices differ from one company to another, especially according to their size and governance model (R1-I1). On the other hand, the experts are unanimous in affirming that teams and operational systems will have to be more autonomous in the future, mainly for agility, responsiveness, and efficiency reasons (R1-I2).

The experts mostly believe that Industry 4.0 technologies will help increase the level of autonomy of operational teams and systems (R1-I3). The answers to Item R1-I4 highlight the strong link between enhancing the decision-making process through Industry 4.0 technologies and increasing the level of autonomy.

The experts agree that it is necessary to distinguish the decision-making process steps that will remain entrusted to people from those where Industry 4.0 technologies are expected to either help people achieve them better or to fully automate them (R2-I1). This choice must consider the risk of rejection of these technologies by employees, the degree of maturity and quality of each Industry technology 4.0, which may be more or less advanced, and the level of mastery and integration of these technologies within each company.

The experts identify a number of risks that can lead to the disempowerment of teams and a loss of autonomy (R1-I3). Many repeatedly point out that the level of autonomy depends heavily on the governance model established in the company (items R1-I3, R1-I4, R1-I5, and R1-I8). This raises the question of whether increased autonomy is a prerequisite for the proper deployment of Industry 4.0 and/or whether it is the deployment of Industry 4.0 technologies—which offers new opportunities through enhanced support for better decision making—that will encourage an increase in the autonomy entrusted to teams.

Some experts stated that using Industry 4.0 technologies to enhance decision making could lead to a strong dependence of operators and managers on these technologies (R1-I4, R1-I6). This could lead to a reduction in the decision-making latitude left to operational teams and the inability of managers to make good decisions in the face of the unknown, or when confronted with problems unforeseen, unmeasured, or difficult to identify by these technologies (R1-I4)

The experts explain that Industry 4.0 technologies can broaden the scope of decision/responsibility entrusted to operational teams (R1-I5). However, some emphasize that this is not an end in itself. Whether or not the decision-making scope of operational teams is broadened will depend above all on the governance model established in the company.

Items R1-I6 and R1-I7 make it possible to assess managers’ expectations as perceived by the panel of experts concerning the first dimension of autonomy [89,91] which is focused on the task.

The experts almost unanimously believe that, overall, Industry 4.0 technologies will enhance the ability of operational teams and systems to carry out their tasks with maximum autonomy. Some explain it by the fact that the automation of certain repeatable and less complex tasks and decisions will relieve operational teams to whom it will then be possible to entrust more complex decisions inducing a higher level of responsibility and autonomy (R1-I6).

Even if a majority of the experts think that this corresponds to an expectation of managers, the increase in the autonomy left to operational teams and systems in the definition of tasks seems to be less consensual (R1-I7) than in the case of task completion (R1-I6). Several experts stress that the growing autonomy of teams must nevertheless take place in “compliance” with standards (R1-I7). In this sense, some experts specify that Industry 4.0 technologies will allow operational teams and systems to evolve standards or better train themselves. However, according to these same experts, the initial definition of standards or the validation of the evolutions retained thereafter will remain the responsibility of managers.

Items R1-I8, R1-I9, and R1-I10 make it possible to assess the managers’ expectations as perceived by the panel of experts concerning the second dimension of autonomy [89,91] focused on the notion of collaboration declined according to three axes: cooperation (R1-I8), communication (R1-I9), and coordination (R1-I10).

On this second dimension of autonomy, the experts’ opinions are the most consensual. The experts widely believe that managers expect Industry 4.0 technologies to enhance the cooperation of operational teams and systems with other organizational entities to increase agility and efficiency (R1-I8). Some issues are highlighted, such as the risk of too much digitalization, the impact of the governance model, or the type of business model (B2B or B2C).

The experts also overwhelmingly believe that managers expect Industry 4.0 technologies to improve communication. They stress that the expectations are strong regarding information sharing and that it will be particularly important to associate meaning with it (R1-I9).

The experts are unanimous that managers expect Industry 4.0 technologies to enable operational teams and systems to more widely manage available resources, synchronize tasks, and align activities (coordination). Some make it a critical success factor. Others point out that the choice of technologies must be aligned with needs that are often unclear or poorly formulated by managers (R1-I10).

Items R1-I11 and R1-I12 make it possible to assess managers’ expectations regarding the third dimension of autonomy [89,91] focused on the notion of governance defined according to two axes: the participation of operational teams in social dialogue (R1-I11) and management style (R1-I12). On this third dimension of autonomy, experts’ opinions are most divided.

No consensus for or against emerges regarding managers’ expectations on the interest of mobilizing Industry 4.0 technologies to allow operational teams and systems to participate widely in social dialogue and promote it. Many point out that this directly depends on the governance model established in the company (R1-I11). Some experts believe that social dialogue must accompany the proper implementation of Industry 4.0 technologies but doubt the existence of a will to exploit these technologies to improve social dialogue. Others point out that they present the risk of promoting unilateral top-down communication.

The experts also appear divided regarding managers’ expectations of Industry 4.0 technologies and their ability to promote the implementation of participative management at the operational level (R1-I12). Many experts stress once again that this depends directly on the governance model established in the company. Some believe that new technologies will offer the opportunity to give more autonomy to operational teams and that this will include implementing a more participative management model. Still, other comments establish links with the directive and persuasive management models.

4.2. Decision-Making Process 4.0: Managers’ Expectations

Overall, managers expect Industry 4.0 technologies to evolve and/or enhance the steps of the decision-making process (cf. Figure 1). Let us consider all the opinions expressed with regard to enhancing the different steps of the decision-making process (items R1-I13 to R1-I20): 77% of the opinions are favorable, 10% are mixed, 8% disagree, and 5% disagree or do not know how to answer.

However, there are strong disparities. The experts are all totally or largely in favor of enhancing the following steps by means of Industry 4.0 technologies:

• Identification of problems and opportunities (R1-I14);
• Diagnosis of problems (R1-I15);
• Real-time “Capture/measure” of information relating to the status and performance of the production system (R1-I13);
• The search for existing solutions (R1-I16); and
• The “evaluation” step of the decision-making process, to ensure that the proposed solution is relevant and adapted to the context (R1-I19).
• On the other hand, the enhancement of three steps is a source of dissensus:
• Filtering and eliminating already known solutions that are only slightly or not appropriate (R1-I18);
• The validation and authorization circuit for the implementation of a chosen solution (R1-I20); and
• The design of new tailor-made solutions (R1-I17).

(The steps are ranked in descending order of consensus in the lists above.)

According to the experts, managers expect Industry 4.0 technologies to contribute to the evolution or enhancement of the first three steps of the decision-making process: “Capture/measure”, “Recognition of a problem or opportunity”, and “Diagnosis”. The experts seem to indicate that these factors correspond to the steps with the highest potential of enhancement for all Industry 4.0 technology groups.

Regarding the “Recognition of a problem or opportunity” step, the experts believe that Industry 4.0 technologies can already offer solutions to enhance the identification of opportunities (and not only problems) through the use of artificial intelligence. However, they point out that this does not systematically correspond to an expectation sustained by all managers. This seems to depend in particular on the degree of maturity of each company in the mastery and knowledge of these technologies (R2-I2).

Then comes the “Evaluation” step, for which the level of expectation of managers and the potential for enhancements are also concerns.

In contrast, according to experts, the “Selection” step is the one for which managers’ expectations are the least important. This seems to be explained by the fact that in an operational context situations are less complex than at the strategic level; if the first steps of the decision-making process are well conducted, then the problem will be properly defined and the number of known solutions available will be relatively small. The mobilization of this “Selection” step is then unnecessary, and it is appropriate to move directly to the “Evaluation” step.

Subsequently, the “Design” and “Authorization” steps are, according to the experts, those for which managers’ expectations are the least important after “Selection”, but also those for which the contribution of Industry 4.0 technologies seems to be the least promising (see Section 4.4).

4.3. Decision Process 4.0 and Level of Integration of Industry 4.0 Principles

The experts point out that companies involved in Industry 4.0 have unequal levels of integration of Industry 4.0 principles (R2-I4). Many companies with a still too superficial understanding of the principles and challenges of Industry 4.0 remain focused on achieving short-term results through the “use of data to better understand reality”.

With respect to several items (R1-I13 to R1-I15, R2-I4, R2-I7, R2-I9, and R2-I11), the experts stress that it is more important to enhance in priority the steps of “Capture/measure”, “Recognition of a problem/opportunity”, and “Diagnosis”, which lay the foundations for the subsequent steps. However, many experts believe that enhancing the first three decision-making process steps remains reserved for companies that have undertaken an in-depth transformation of their operational governance mode (R2-I4). In particular, they draw attention to the importance of empowering field teams to use these new technologies and associated systems so that they can participate in their evolution and improvement.

Even if some technologies, such as artificial intelligence, offer great potential for improvement, experts generally agree that the majority of manufacturers are not at a sufficient level to consider today the enhancement of the step “Search” for already known solutions, especially in SMEs (R2-I5).

Similarly, and according to the experts, if the enhancement of the “Selection” step does not appear to be a priority today in the eyes of managers (R1-I18), this may change in the future (R2-I3). This is explained by the often too limited number of known solutions currently available and mobilizable within many companies. However, the deployment of Industry 4.0 technologies will enhance the ability to identify solutions and remember them for next time. The number of known solutions is therefore likely to grow. AI also makes it possible to:

• Improve the quality of the selection of a known solution even within a limited panel of candidate solutions; and
• Identify a greater number of known possible solutions by recognizing similarities between a priori unrelated situations.

The experts are almost unanimous in saying that most manufacturers have not yet reached a sufficient level of integration of the principles of Industry 4.0 to consider today the enhancement of the “Design” step (R2-I6). Several experts point out that this step will probably be where human intervention will remain the most essential.

According to the experts, the same applies to the enhancement of the “Evaluation” step (R2-I7), but the positions here are much less clear-cut than before. This step nevertheless is the locus of highest consensus with respect to enhancement after the first three steps of the decision-making process, particularly via the use of simulation, performance measurement, or automated reporting systems. Accordingly, if the level of integration of the principles of Industry 4.0 is not yet perceived as sufficient by the panel of experts, expectations remain high (R1-I19), and partial solutions already exist to enhance this step.

Finally, the fact that the actual level of adoption of Industry 4.0 principles is too low to consider the enhancement of the “Authorization” step is also almost unanimous (R2-I8).

The experts remain relatively divided on the future orientations to be given around the enhancement of the “Authorization” step in terms of level of delegation or automation (R2-I9 to R2-I11).

The experts generally remain reluctant to fully automate the “Authorization” step (R2-I11). They are concerned with the risk of disempowerment of managers, loss of initiative of operational teams, and inadequate management systems in place for which the interpersonal component is central (nemawashi, for example, which is the process of consensus-building in a Lean management context [92].

In this sense, several experts believe that it is worthwhile to mobilize Industry 4.0 technologies to facilitate communication and coordination (R2-I9) to make the “Authorization” step more efficient and reduce the implementation time of the validation loop. However, in a more consensual way, the panel of experts thinks that it is better to reinforce the last step “Authorization” to partially or totally delegate this step from the manager to the team or person at the initiative of the proposed solution (R2-I10).

4.4. Enhancement of the Decision-Making Process through Industry 4.0 Technologies

Items R1-I21 to R1-I27 and R2-I24 to R2-I26 aimed to target the steps of the decision-making process likely to evolve or be enhanced by the ten Industry 4.0 technology groups: big data analysis, artificial intelligence, Internet of Things (IoT), simulation, augmented reality, cybersecurity, cloud computing, cyber-physical systems, autonomous robots/machines and inter-machine communication (M2M).

The summary of the results appears Figure 3 below.

Figure 3, above, shows the percentage of experts who believe that one of the steps in the decision-making process listed in the row is likely to evolve or be enhanced by one of the Industry technologies 4.0 listed in the column.

For example, 71% of experts believe that the “Capture/measure” step is likely to evolve or be enhanced by big data analysis:

• Cells corresponding to a “consensus for” appear in green in the Figure 3 (i.e., responses in agreement with the association of a technology with the evolution or enhancement of one of the steps of the decision-making process ≥60% of all opinions);
• Cells corresponding to a “consensus against” appear in red in the Figure 3 (i.e., responses disagreeing with the association of a technology with the evolution or enhancement of one of the steps of the decision-making process ≥60% of all opinions); and
• Cells corresponding to a “dissensus” appear in white.

The next column gives an indication of the steps in the decision-making process that are most likely to evolve or be enhanced by one of the Industry 4.0 technologies listed in the column. For example, 16.7% of all possibilities for evolution or enhancement of the decision-making process steps by one of the ten Industry 4.0 technology groups listed in the column concern the “Capture/measure” step.

The last two columns make it possible to compare, based on the opinions formulated by the experts, the level of expectation of managers and the potential offered by the ten Industry 4.0 technology groups to evolve or enhance the steps of the decision-making process.

The three penultimate lines indicate the number of experts:

• Who believe that one of the technologies listed does not enhance any of the steps in the decision-making process;
• Who don’t know how to answer; and
• Who do not want to answer.

Finally, the last line of the Figure 3 indicates the technologies most likely to develop or enhance all or part of the steps of the decision-making process. For example, 12% of all opportunities for change or enhancement of the decision-making process steps by one of the ten Industry 4.0 technology groups listed in the column come from big data analysis.

4.4.1. Contribution of Industry 4.0 Technologies to Enhance the Decision-Making Process

The experts stated that the Cloud contributes to the evolution and/or enhancement of all steps of the decision-making process by promoting the pooling and sharing of information and collaboration logic (R1-I27).

The experts’ opinions seem to indicate that among the nine remaining technological groups (Figure 3), big data analysis and artificial intelligence are the technologies likely to significantly evolve or enhance the largest number of steps of the decision-making process (R1-I21 and R1-I22). According to the results, these two technologies can help to enhance similar steps in the decision-making process, except the “Capture/measure” step. The experts believe that these are particularly promising technologies to enhance:

• The first steps of “Recognition of a problem or opportunity” and “Diagnosis” of the decision-making process; and
• The “Search for solutions” and “Selection” steps specific to situations for which solutions are already known.

On the other hand, dissensus remains concerning the enhancement of the “Evaluation” step by these two technologies.

The experts mostly agree that inter-machine communication (M2M) is likely to mainly enhance the first three steps “Capture/measure”, “Recognition of a problem or opportunity”, and “Diagnosis” (R1-I30 and R2-I26).

The experts prioritize the use of cyber-physical systems to enhance the first steps “Capture/measure” and “Recognition of a problem or opportunity” (R2-I24). However, several experts believe it has become an “umbrella term” and a concept perceived as very vague. They specify that these systems result from the amalgamation of several technologies and that they generally integrate themselves into other systems (R1-I28 and R2-I24). Their direct contribution relates more naturally to enhancing the first steps of the decision-making process. Still, they can indirectly contribute to enhancing all the steps of the decision-making process (R2-I24).

The experts’ answers indicate that autonomous robots/machines mainly offer opportunities for enhancement of the first two steps of the decision-making process, “Capture/measure” and “Recognition of a problem or opportunity” (R2-I25). The majority of experts agree that they will contribute little to the direct enhancement of the last six steps of the decision-making process and remain mainly actuators capable of capturing data and communicating with other systems (R1-I29 and R2-I25).

Although often mentioned by experts, “augmented reality” appears to be one of the technologies most suitable for dissensus (R1-I25). Expert opinions seem to indicate that this technology group is not conducive to evolving or enhancing the “Capture/measure”, “Search for already known solutions”, “Selection”, and “Authorization” steps. The experts remain undecided or divided as to the relevance of this technological group to enhance the steps “Recognition of a problem or an opportunity”, “Diagnosis”, “Design”, and “Evaluation” (see Section 4.4.2).

In an extremely consensual way, experts believe that the IoT is reserved for enhancing the “Capture/measure” step but does not relate directly to the other steps of the decision-making process (R1-I23).

The experts indicate that simulation systems mainly contribute to enhancing the “Evaluation” step of the decision-making process to assess solutions’ relevance and choose a solution that can be judiciously implemented. However, simulation seems relevant to enhancing the “Design” step, allowing the modelling of several innovative solutions before selecting the most promising one (R1-I24). The same applies to the “Selection” step, but the experts have reservations about the interest of enhancing this step in the short term (cf. Section 4.3—item R2-I3).

Regarding the enhancement of the different steps by technologies related to “Cybersecurity”, the opinions of experts are quite divided (R1-I26, R2-I22, and R2-I23):

• The majority believe that cybersecurity should protect the exchanges between the various stakeholders at the last step “Authorization” before the action is taken;
• A number of them believe that these technologies must make all the steps of the decision-making process more reliable and not just one or a few particular steps. They insist that if the slightest link is corrupted, the entire decision-making chain is corrupted (R2-I22 and R2-I23). These experts believe that “Cybersecurity” necessarily accompanies any step in which data is generated, exchanged, transformed, interpreted, or stored (R1-I26); and
• Others believe that it is advisable to act as a priority at the beginning of the process on capture/measurement so as not to work from corrupted data. Reference is often made here to the expression “garbage in, garbage out” (R2-I22), which refers in computer science to the concept that erroneous or absurd input data (garbage) produces absurd results. However, one of the experts draws attention to the risk of confusion between cybersecurity and corruption of data entering the decision-making process. It considers that this last point is not in the field of cybersecurity but depends more on the quality and robustness of the steering and management processes.

4.4.2. Enhancement of the Steps of the Decision-Making Process

The vast majority of experts agree to highlight the complementarity of IoT and big data analysis to enhance the “Capture/measure” step (R2-I12). The IoT appears as the preferred technology to “capture” real-time data coming back from the field (R1-I23) even if it can be helped in this by cyber-physical systems, inter-machine communication (M2M), and data from autonomous robots/machines (R2-I24 to R2-I26). Several experts specify that the analysis of big data not only contributes to a simple “measurement” of indicators reflecting the state and performance of the production system, but also allows the identification of links between the variables measured and a predictive approach (R2-I12). The experts point out that big data analysis often works in tandem with artificial intelligence, although some experts believe that the latter intervenes rather than strengthens the next steps of the decision-making process (R2-I13).

At the level of the “Recognition of a problem or an opportunity” step, half of the experts agree that the use of augmented reality is mainly considered to alert an operator to a critical situation to which he does not have direct access or that he may not spot (R2-I15). They explain that augmented reality can be used to reveal problems or opportunities by:

• Enriching what is perceived by the operator; and
• Facilitating information sharing between a remote expert and a worker/operator physically present on the workstation.

Other experts are more divided on this point and object that simpler solutions to implement are generally possible and preferable. However, this technology seems promising to enhance this step for training or demonstrations.

At the “Diagnostic” step:

• A large majority of experts agree that the use of simulation is mainly considered to evaluate the possible consequences of a problem on a future state of the operational system or to estimate the potential offered by an opportunity (R2-I16); and
• The majority of experts believe that augmented reality must be coupled with simulation to facilitate the visualization and understanding of the current or future state of the operational system. However, some point out that augmented reality involves great efforts in terms of development and that simulation alone is often sufficient (R2-I17).

The experts explain that cloud computing, big data analysis, and/or artificial intelligence offer interesting opportunities to enhance the “Search for already known solutions” step. However, some experts explain that many managers say they are interested in this possibility of enhancement but do not make it a priority today (R1-I16), in particular for the following reasons:

• The volume of data and the capitalization of knowledge are still too low and/or poorly organized, especially regarding the root causes of each known problem and the solutions that can be brought to it. In this sense, this justifies the prioritization given to the enhancement of the first three steps of the decision-making process (R1-I16);
• Many managers do not believe in the exploitation of technologies such as artificial intelligence to process this data (R1-I16). Yet some experts explain that the use of techniques such as neural networks or case-based reasoning, fuzzy logic, or a combination of these techniques to sort or identify similarities between a priori unrelated situations can already prove useful even if the number of known solutions is relatively small (R2-I3); and
• When there are several known solutions (even in very limited numbers), the reflex is usually to choose the one that has worked best in the past without resorting to a detailed analysis aimed at comparing the different situations and choosing the best solution taking into account the specificities of each situation. This is generally justified by the limited time available to carry out this analysis (R2-I3).

The coupling of cloud computing, big data analysis, and artificial intelligence seem to offer interesting enhancement opportunities for the “Selection” step. However, the experts’ comments indicate that the situations encountered at the operational level within companies are currently not likely to be able to establish a link with a large number of solutions already known (R1-I18). Therefore, the “Selection” step of the decision-making process seems today to be reserved for decisions at a more tactical or strategic level. Many experts believe that this will change later when the integration of Industry 4.0 will be more effective (cf. Section 4.3—item R2-I3).

The majority of experts consider that the two most promising technologies to evolve or enhance the “Design” step are simulation and augmented reality (R2-I18). Many experts who agree with this statement point out in their comments that priority should generally be given to simulation to explore new scenarios and solutions and then test them in a pilot, possibly using augmented reality and, more precisely, virtual or mixed reality. It should be noted that some experts believe that the role that AI could play in enhancing this step is underestimated.

At the level of the “Evaluation” step, the analysis of the opinions formulated by the experts makes it possible to identify the following points:

• The coupling between artificial intelligence and simulation is not systematically necessary; simulation systems can be operated without AI (R2-I20). However, this remains the most widely promoted coupling by the panel of experts (R2-I19 to R2-I21);
• The coupling between big data analysis and simulation is not systematically necessary, mainly because the implementation of simulation systems does not always require the use of a very large amount of data (R2-I19);
• Even more markedly, the coupling between augmented reality and simulation does not appear to be systematically necessary (R2-I21);
• In some cases, augmented reality can complement simulation systems to facilitate the visualization of the consequences and results of the scenarios and solutions envisaged. However, a large number of experts believe that priority should generally be given to simulation in order to test a solution envisaged (R2-I21);
• Artificial intelligence can be used, for example, to detect different patterns in datasets and thus highlight relationships or impacts between solutions and problems without going through simulation (R2-I20); and
• Finally, an expert insists that coupling with other technologies can usefully contribute to the enhancement of this “Evaluation” step and, in particular, augmented reality (R2-I19 and R2-I20).

As explained above (cf. Section 4.2), the experts remain relatively divided on the future guidelines to be given in the enhancement of the “Authorization” step (R2-I9 to R2-I11). Some experts agree that it is interesting to mobilize Industry 4.0 technologies to facilitate communication and coordination (R2-I9) to make this step more efficient and reduce the time to implement the validation circuit. In this sense, cloud computing and cybersecurity appear as the technologies with the highest potential for enhancements (R1-I26 and R1-I27). However, the strategies for enhancing this step will depend on the governance model in place (R1-I1 and R1-I12) and the autonomy 4.0 model targeted in the long term by each company.

5. Research Results, Limitations, and Perspectives

5.1. Main Results

This research work highlights the technology groups that are most conducive to enhancing each step of the decision process. Figure 4 summarizes the contributions of Industry 4.0 technologies to decision-process enhancement that are in consensus (cf. solid line link on Figure 4) and the contributions that are under debate (cf. dashed links on Figure 4). The latter needs to be clarified in the future as the level of integration of Industry 4.0 principles increases within companies and as the capabilities offered by Industry 4.0 technologies evolve or become more precise.

The synthesis of the expert comments highlights that a technology group alone can rarely contribute to enhancing a decision step. On the other hand, a given group of technologies may be useful in enhancing one or more steps of the decision process for a given type of a decision but be inappropriate in another case. The complexity of the problem, the nature of the data needed to identify and analyze the situation, the number of known solutions and their characterizations, the evaluation of the implementation of a solution, and the decision-making circuit leading to the implementation of the chosen solution can be very different from one type of decision to another and from one company to another.

Therefore, it seems difficult to identify combinations of technology groups that would be systematically relevant and generalizable to enhance all or part of the decision process for any type of decision and industrial context.

However, some remarkable points concerning each technology group emerge from the study. Cloud computing occupies a special place by potentially contributing to enhancing all decision-making process steps. This technology appears to act as the backbone to any system of global enhancement of the decision-making process. Conversely, the IoT offers potential for enhancement solely focused on the “Capture/measure” step. However, it appears to be an essential and decisive technology for enhancing this step. The autonomous robots/machines and inter-machine communication technologies groups have a marked potential for enhancing the early steps of the decision-making process. Still, their contribution seems much less interesting beyond the “Diagnosis” step. The contribution of cyber-physical systems seems particularly interesting for enhancing the first steps of the decision-making process. Still, this technological group also seems to be able to contribute indirectly to enhancing all the steps of the decision-making process. Simulation systems offer the potential for enhancements focused primarily on the “Evaluation” step. In the long term, this technological group could also play a more marked role in selecting known solutions, designing tailor-made solutions, or even at the “Diagnosis” step. Big data analysis and artificial intelligence are two technological groups whose contributions appear to be closely linked. Their role seems particularly promising in enhancing the first three steps of the decision-making process and searching for and selecting known solutions. They also seem to play an interesting role in the long term to enhance the “Evaluation” step. Their interest seems less obvious today in the case of tailor-made solutions, but this could evolve in the long term depending on future progress in the development of these technologies and the level of integration of the principles of Industry 4.0 within companies. The contribution of augmented reality is less generalizable and seems to be reserved for specific application cases, especially in addition to other technologies for the enhancement of the steps “Recognition of a problem or an opportunity”, “Diagnosis”, “Design”, and “Evaluation”. While the potential for cybersecurity to enhance the last “Authorization” step seems to be established, there is some dissensus on the enhancement of the other steps with two positions displayed:

• Cybersecurity must act primarily at the beginning of the decision-making process on capture/measurement to avoid working with corrupted data; and
• This technology group must make all the steps of the decision-making process reliable because if the slightest link is corrupted, the entire decision-making process is corrupted.

These different points of view call attention to the need to clarify and communicate widely on the exact scope covered by cybersecurity, particularly on the role played by this technological group concerning the corruption of data used in the decision-making process. The same applies to the field covered by cyber-physical systems, which is often perceived as very vague and associated with implementations of very different levels of aggregation.

5.2. Limitations

The first limitation of this study concerns the Delphi method and the panel of experts selected by the steering committee. The choice of experts necessarily involves a subjective bias related to the choice of experts in academic and industrial networks. However, despite the steering committee’s desire for neutrality and the particular attention paid to the coherence and complementarity of the profiles selected in relation to the research objective, it nevertheless seems difficult to completely free oneself from these difficulties.

A specific limitation concerns the responses to the items on managers’ expectations. Indeed, they only reflect the perception that the panel experts have of these expectations. Despite the precautions taken in the selection of “industrial” and “integrator” experts, it would be interesting to carry out a specific study on the differences between the expectations formulated by managers and “practitioners” and the perception of these expectations by the decision-makers in charge of Industry 4.0 deployment projects within companies

Another limitation of this study is the design of the first questionnaire, which plays a decisive role in the orientation given to the study and in the understanding and appropriation of the subject by the panel of experts. It was built around the research objective and decision-making 4.0 model in an operational context proposed by Rosin et al. (Figure 1). It was then corrected to consider the feedback and expectations formulated by a test panel different from the panel of experts who then participated in the entire study. The framework note attached to the questionnaire could also be improved accordingly. Although this intermediate step has reduced the risk of misinterpretation due to problems in formulating the various items, it cannot be eliminated. In all cases, the resulting choices inevitably had a structural implication at the beginning of the study. However, experts were allowed to comment and propose other statements in the following. The orientation of the first questionnaire likely created a subjective bias from the outset.

The last limitation is that our study is qualitative, the Régnier abacus having been specifically adapted to this type of research. However, the prospective aspect of the subject of study and its scope did not allow the current state of knowledge to envisage a more quantitative approach. However, the Delphi–Régnier method allowed us to collect empirical data and visualize experts’ opinions to identify the main contributions of the different technological groups to enhance the decision-making process. It also provided the basis for constructing a more quantitative study on managers’ current and future expectations on this subject. Finally, it also opens the field to a comparative qualitative study on the new modes of autonomy 4.0 that enhance the decision-making process.

5.3. Research Perspectives

The study highlights more research opportunities. It does not pronounce on whether or not to extend the decision-making remit entrusted to operational teams and systems. The authors believe that research should be undertaken to study the evolution of this decision-making remit that may be induced by the enhancement of the decision-making process by Industry 4.0 technologies.

On several occasions, a link is implicitly established with certain principles and notions of Lean, such as standards, continuous improvement, gemba, and nemawashi. Lean approaches often appear as a prerequisite for the proper deployment of Industry 4.0 [93]. Additionally, it would seem particularly interesting to study to what extent the enhancement of the decision-making process by Industry 4.0 technologies is naturally consistent or conversely risks contradicting the implementation of the different principles of Lean.

Finally, this study highlights the close link between enhancing the decision-making process and the evolution of the level of autonomy that can be entrusted to operational teams and systems. The authors believe that this calls for additional work on new modes of cyber-autonomy, risks, opportunities, and associated critical success factors and their implementation. The enhancement of the decision-making process inevitably raises questions about the advisability of a new distribution of responsibilities entrusted to the operational teams and systems and about the new modes of autonomy induced by the deployment of Industry 4.0. Today, little work is interested in the expectations of managers on these points and these fundamental links that are likely to upset the models of organization at work. Previous work by Rosin et al. [72] has led to the proposed model of autonomy types composed of seven types of autonomy in the decision-making process based on Industry 4.0 technologies: (1) Cyber Monitoring, (2) Cyber Search, (3) Standard Decision Support, (4) Cyber Control, (5) Cyber Design, (6) Customized Decision Support, and (7) Cyber Autonomy. This model has been validated by analyzing a set of case studies from the literature but must now be validated with real cases and feedback from business experts.

6. Conclusions

The research work presented in this article aimed to study the potential to enhance the decision-making process of operational teams and systems with the use of Industry 4.0 technologies. As such, this study clarified the possible contributions of the various Industry 4.0 technology groups to the enhancement of the decision process and identified the technologies with the greatest potential for enhancement at each step of the decision-making process. This study’s results can help decision makers target the technologies to be considered in priority according to the type of decision process reinforcement desired.

This study has made it possible to specify the decision-making process steps for which managers expect an enhancement by Industry 4.0 technologies. According to the expert panel, managers’ expectations today focus mainly on enhancing the steps of the decision-making process if already known solutions can be identified and proposed: “Capture/measure”, “Identification of problems and opportunities”, “Diagnosis”, “Recovery of existing solutions”, and “Evaluation”. The enhancement of the first three steps is expected as a priority, as it lays the foundations for improving the next steps. Conversely, the level of integration of the principles of Industry 4.0 is not currently considered sufficient to consider the enhancement of other steps. Enhancing the “Design” step appears to be the least challenging in the short term, as it is generally considered very complex. This is also the case for the “Selection” step, particularly because of the low capitalization of data on known problems and solutions. The expectations around the enhancement of the “Authorization” step are the ones that lend themselves most to the discussion because they are directly related to the governance model and the degree of autonomy targeted within each company, as well as to the risks induced by these choices. This study confirms the strong link between improving the decision-making process through the use of Industry 4.0 technologies and increasing the level of autonomy. While it appears that Industry 4.0 technologies will increase the level of autonomy of operational teams and systems, there are still many risks that can lead to a loss of autonomy or disempowerment of operational teams and managers. The results of this study highlight that managers particularly expect Industry 4.0 technologies to contribute to strengthening the level of autonomy centered on the notion of collaboration declined according to three axes: cooperation, communication, and coordination. The enhancement by Industry 4.0 technologies of task-centered autonomy also seems expected by managers. However, the results are more ambiguous about the autonomy left to operational teams in defining their tasks. This seems to depend heavily on the governance and management model implemented and the level of integration of Industry 4.0 principles within the company. In this sense, managers’ expectations according to the expert panel seem to lead to strong dissensus at the level of enhancement by Industry 4.0 technologies with respect to the level of autonomy focused on governance.