Abstract

Industry 4.0 promises better control of the overall product development process; however, there is a lack of computational frameworks that can inject human factors engineering principles early in the design. This shortage is particularly crucial for prototyping human-centered products where the stakes are high. Thus, a smooth Industry 4.0 transformation requires keeping ergonomics in the loop, specifically to address the needs in the digitized prototyping process. In this paper, we explore a computational prototyping approach that focuses on various fidelity levels and different human–product interaction levels when conducting ergonomics assessments. Three computational prototyping strategies were explored, including (1) a digital sketchpad-based tool, (2) computer-aided design and digital human modeling-based approach, and (3) a combination of computer-aided design, digital human modeling, and surrogate modeling. These strategies are applied to six case studies to perform various ergonomics assessments (reach, vision, and lower-back). The results from this study show that the designers need to consider the tradeoffs between the accuracy of ergonomic outcomes and resource availability when determining the fidelity level of prototypes. Understanding the intricacies between the fidelity level, type of ergonomic assessment, and human–product interaction level helps designers in getting one step closer to digitizing human-centered prototyping and meeting Industry 4.0 objectives.

1 Introduction

The engineering community is in the midst of a rapid transformation of how the product is designed and manufactured. Over the last few years, significant advancements in computers, sensors, and communications technologies have accelerated hyper-connected smart design and manufacturing systems. With this shift, engineering practices have moved towards considering the entire product life-cycle in addition to production aspects during the manufacturing process [1]. At the heart of this technology-driven transformation, digitization via the exponentially growing technologies (e.g., internet of things (IoT), automation, additive manufacturing) are paving the way for cyber-physical systems design-merging the real and virtual worlds.

The above trend is often regarded as Industry 4.0 and has been used across Europe, particularly in Germany’s manufacturing industry since around 2010. Around the same time, North American industries adopted a similar production practice called “Industrial Internet” [2]. Likewise, France and China introduced “Industrie du futur” and “Made in China 2025,” respectively [3,4]. Today, Industry 4.0 is both an inspiration and opportunity for global competitiveness [5,6] as many refer to it as the fourth industrial revolution. Thus, it is expected that Industry 4.0 will bring a significant shift in the industry by incorporating digitization of production, automation, and linking manufacturing plants with supply chain [79]. The definitions, key concepts, core technologies, and ways of implementing Industry 4.0 practices are continuously evolving [10,11]. The Industry 4.0 needs an advanced architecture that is highly technologically complex, which is why it is yet to reach maturity in terms of solving broader technical problems [12,13]. Industry 4.0 requires further research on the theories and implementations from multidisciplinary domains such as customization, optimization, automation, decisions support, human–machine interaction, and digitization [9,11,14,15]. In addition to studying the broader scope, it is also vital for engineers to explore the building blocks or particular elements (cogs in the wheel) that make up the whole to meet Industry 4.0 goals [16]. One such cog that has a critical role in designing modern products is the computational models that enable better early design prototypes.

Prototyping, an important phase of product design, is known to be resource-hungry (time, cost, material, machines, and personnel) [17]. As a result, prototyping significantly impacts the overall production [17]. This paper introduces a computational prototyping study where the fidelity levels and human–product interaction levels are explored to identify their effects in prototyping human-centered products. Identifying and understanding the intricacies between the fidelity, interaction level, and ergonomic assessment will help to build an effective computational prototyping approach that can be a key facilitator to meeting Industry 4.0 goals. The computational prototyping approach highlights the centrality of design by injecting computational ergonomics workflows that allow designers to capture human–product interaction-related issues early in design before functional prototypes are built. This approach overlaps with the premises discussed within Industry 4.0 by focusing on customer-oriented mass customization, simulating human–machine interactions early in design, and injecting human factors throughout the value-chain, thus improving the efficiency of product development [16,18,19]. The human-centered prototyping approach discussed in this paper has the potential to aid design companies in reducing the overall design time and cost and improving other factors such as quality, risk, and overall environmental sustainability. The main contribution of the paper is to build a building block for a computational prototyping approach focusing on injecting ergonomics in human-centered products. This computational approach will help in the digitization, consideration of ergonomics, and in the overall product design process in the age of Industry 4.0.

The road-map of the paper is as follows: Sec. 2 presents a literature review on the building blocks of the prototyping methodology: (a) prototyping in the human-centered design (HCD) domain and (b) digital human modeling (DHM). Section 3 talks about the prototyping methodology and presents a case study for illustrating the research objectives. Sections 4 and 5 contain results and discussions, and Sec. 6 wraps up the study by summarizing the limitations and future work.

2 Background

Many studies have shown that injecting ergonomics early in design (a proactive approach before risky events occur) enables designers to implement human factors engineering (HFE) guidelines better to mitigate potential risks, allowing the development of ergonomically sound products or workplaces [20]. Hence, the proactive ergonomics approach provides designers a better strategy to develop products that are improved in quality by encompassing human–product interactions as an essential part of the continuous improvement process, not as a one-time event. Also, eliminating retrofitted design changes decreases the lead-time to market and requires fewer resources [21].

Incorporating HFE guidelines during product development involves collecting human–product interactions data, which is often not widely available [22]. Alternatively, designers can utilize prototyping as a method to simulate human–product interaction by creating either a physical prototype, computational prototype, or mixed prototype. Physical prototypes are advantageous in representing form and functionality; however, they are time-consuming and costly to build [23,24]. In contrast, computational prototypes are low-cost and faster to develop; however, they lack the fidelity in representing physical interactions between human operators and products, limiting the extent of feedback [25,26]. Another concern during prototyping is the level of interaction between the user and the product. Duffy mentioned that a physical prototype would be a better choice if there is a high-level interaction between the user and the product. In contrast, when a low-level interaction exists, a computational prototype is preferable [27].

Depending on the level of interactions (e.g., from low to high) between the users and products, designers need to agree on the type (physical or virtual), fidelity (low or high), and complexity (low or high) of the prototypes before the embodiment phase starts. However, there is a lack of understanding and guidelines on systematic prototyping solutions that can help designers navigate the above considerations. This work aims to explore the intricacies between the fidelity level, human–product interaction, and ergonomic assessment level in a computational prototyping approach [28]. This association can support designers in creating more effective prototypes for human-centered products to evaluate ergonomics in the early phases of the design process. The overall approach also supports the Industry 4.0 objectives, within the scope of the cyber-physical systems, by utilizing sensors and virtual reality data to inform digital twin driven early design ergonomics decision-making.

In this paper, human–product interaction is defined by borrowing the concepts from human–computer interaction and human–robot interaction. The interaction between the human and system is regarded as actions that the human operators perform to a system and the feedback that the operator receives from the system. The interactions in this context can be either complex or simple, depending on the number of actions and feedback that are present. In this research, three computational prototypes of low-, mid-, and high-fidelity levels are used to prototype and evaluate the ergonomic assessment of six conceptual products having low, mid, and high human–product interaction levels.

2.1 Prototyping in Human-Centered Design.

In the HCD domain, one of the critical aspects of employing prototyping activities is to detect HFE design issues that can negatively affect the product’s overall performance and the well-being of the user. One can see in the literature that various attempts have been made to develop prototyping taxonomies that aid designers in planning prototyping strategies systematically. Multiple taxonomies have been classified in terms of cost, stage of design, level of abstraction or realism, and intended evaluation purpose [29,30]. Prototyping classification has also been made based on the process used to create a prototype, such as material removal or material addition [31]. One of the shortcomings in these classification approaches is the lack of broad coverage of the prototyping design space [32]. A more comprehensive taxonomies of prototyping, which was developed based on variety (physical or computational), complexity (system or component), and fidelity (high or low) [3234], served as the foundation for the prototyping studies used in this paper.

There are various advantages and disadvantages for each type of prototype. Physical prototyping is an effective strategy when designers want to evaluate the shape composition and functionality of a product [23,35]. Three-dimensional (3D) physical prototypes are best at representing the shape relations and providing visual and tactile feedback [36]. In traditional ergonomics studies, the standard industrial practices involve building physical prototypes and conducting human subjects experiments to evaluate operator performance [37]. However, physical prototypes take a long time to build; they are inflexible to modifications and costly [24]. Alternatively, computational prototypes can be built quicker. They are easier to share and transfer between different parties involved in the design process because there are no shipping and handling concerns. Additionally, a computational prototype’s flexibility allows it to be used repetitively without creating a new prototype every time a design change is made [25]. These characteristics enable computational prototypes to be built faster in a cheaper way and used earlier in the design process. However, the lack of physical and sensory attributes, such as haptic and olfactory and the absence of the sensation of weight, limit the fidelity of representing human–product interactions [26].

Besides deciding whether to build a physical or computational prototype, designers need to determine whether to develop a low- or high-fidelity prototype. There are perplexing views in the literature regarding the appropriate level of fidelity [3840]. For example, a study focusing on evaluating a lighting controller interface showed that a high-fidelity prototype reveals twice as many design problems compared to a low-fidelity prototype [40,41]. High-fidelity prototyping provides richer sensory feedback and a higher level of interaction compared to low-fidelity prototypes. These attributes facilitate the identification of more design problems but at the expense of higher cost and development time [40]. In contrast, examples taken from numerous design studies show that low-fidelity prototypes are preferred over high-fidelity prototypes when modeling fundamental design attributes because the low-fidelity prototypes are easy, quick, and cheaper to build [36]. Some other studies claim that low-fidelity and high-fidelity prototypes are equally suitable in evaluating usability issues in interface design [39,42].

In summary, there are various strategies to define a suitable prototyping medium to evaluate human–product interactions. One key take is that designers need to consider multi-faceted factors such as the type (physical or computational), fidelity level (high or low), interaction level (high or low), cost, and time spent on developing a prototype. Overall, as the number of factors increases, selecting the correct prototyping strategy, particularly for the design of human-centered products, becomes a perplexing query. The literature review shows that different studies offer contrasting results and views. Thus, there is a lack of comprehensive and widely accepted guidelines for designers to follow on building prototypes early in design [43].

2.2 Digital Human Modeling for Ergonomics.

Digital human modeling (DHM) is a computational prototyping approach used for evaluating the ergonomics of products and workplaces. Commercially available DHM software has graphical representations of humans (manikins) with mathematics and science in the background [44,45]. DHM software can import computer-aided design (CAD) representations of products and workplaces to facilitate the prediction of injury and performance. There are several DHM software commercially available in the market such as ramsis [46], santos [47], delmia [48], and jack [49], which have interfaces that allow importing CAD models. Ergonomics analysis modules built within DHM software range from biomechanical analysis for manual material handling tasks to vision analysis for vehicle operations and time studies for assembly planning to energy expenditure and fatigue assessments for workers performance measurements [50]. Additional information about DHM software can be found in Refs. [45,50].

The use of DHM as a design support tool within engineering design ranges broadly. Often it is used for assessing concept products to discover ergonomics issues. For example, Colombo et al. performed an ergonomic analysis of a family of three refrigerator products. It was reported that different users had different reaching and vision performances as they interact with the refrigerator. For instance, some areas on the fridge were accessible to 95th percentile population, and some postures during maintenance did not conform with the National Institute for Occupational Safety and Health (NIOSH)’s lifting index, which increases the risk of injury [51]. In another example, DHM was used by the Ford Motor Company to find out the minimum clearance between potential drivers and the automobile interior panels. Designers performed a swept volume analysis to evaluate the minimum clearance for interior designs [52]. Likewise, DHM is also heavily used in the aviation industry, both military and civilian projects. For example, during the development of the F-15 fighter jet, DHM was used to assess whether a technician can reach and pull a heavy object during the installation of the radar equipment. Reach envelope and static strength analyses were performed to generate ergonomic reviews [53]. DHM is also used in healthcare [54], space research [55], sports [56], manual assembly [57], and consumer product development domains [58].

3 Methodology

In a previous study, different types of prototypes are explored to identify the suitable prototype to assess human performance during emergencies in a cockpit [59]. This study is a continuation of the methodology presented in that work [28]. This research focuses on identifying the appropriate level of fidelity for computational prototyping with varying degrees of human–product interactions. Figure 1 presents three computational prototyping methodologies with varying human–product interactions and fidelities, which are highlighted with rectangular boxes. Prototyping Method #1, a two-dimensional (2D) online sketching tool, has the lowest fidelity in terms of product visualization and ergonomics analysis capabilities. The integration of CAD and DHM represents the prototyping Method #2, which forms a digital prototyping environment that enables designers to perform quick ergonomic assessments based on three-dimensional (3D) CAD models and DHM ergonomics toolkits. Finally, the integration of CAD, DHM, and surrogate modeling is referred to as Method #3, which has the highest level of fidelity among the three methods described in this research due to its human performance and safety optimization capabilities. The prototyping methodologies are explored to understand how different fidelity levels with varying levels of human–product interactions affect the early design prototyping efforts. The following sections provide more details about each prototyping method.

3.1 Prototyping Method #1: Digital Sketchpad.

Sketchpad 5.1, an online sketching tool [60], is chosen as a method to represent a low-fidelity prototyping methodology. Sketchpad 5.1 is a two-dimensional (2D) digital sketching pad where designers are given a blank canvas with drawing tools and stencils to conceptualize ideas via freehand style sketching. Unlike the DHM, Sketchpad 5.1 has no integrated anthropomorphic database to assist in creating the human form and no algorithms to assign postures. Instead, designers need to use anthropomorphic charts, ergonomics guidelines, or online databases to represent body-proportions as stick-figures. Furthermore, designers can use stick-figures with geometry relations to perform quick-and-dirty ergonomic evaluations (e.g., 2D reach volume). However, even with anthropomorphic guidelines and geometric relationships, this prototyping method has the lowest fidelity in terms of its ability to mimic ergonomics of actual product use scenarios. The complexity of analyses highly depends on the expertise and human factors knowledge of the designer. Also, the 2D nature of the sketching interface adds to its limitations. Still, this method is used in early product development stages, especially during ideation and product conceptualization. It is significantly faster and less resource-intensive when compared to computationally expensive DHM models.

3.2 Prototyping Method #2: CAD and DHM.

Prototyping Method #2 uses CAD software to create product/workplace geometry and DHM to execute ergonomics analysis. In this study, the CAD file of the product is exported to Siemens jack, [49] a DHM software, to conduct ergonomic evaluations. In terms of its ergonomics evaluation capabilities, this method has a higher fidelity than prototyping Method #1. DHM software includes anthropomorphic databases and inverse kinematic toolkits, which help designers to create manikins and assign realistic postures and motions. Also, various types of ergonomic analyses such as reach zones, vision obscuration, and lower-back compression assessments can be performed without conducting physical experiments. This method can be used to evaluate the ergonomics of products with low to intermediate complexity. However, if the aim is to design a product that has high levels of human interactions, this method has some limitations. For example, one of the fundamental issues is that designers need to know the design variables that affect human performance beforehand. In the absence of this information, designers explore numerous configurations and investigate many options before reaching a consensus. Thus, this approach requires the exploration of the entire design space. Prototyping Method #2 does not facilitate any computational tools for designers to explore the design space and determine the optimal human performance. With the absence of optimization methods, designers often rely on personal expertise and develop subjective assessments to explore potential design configurations. This approach often includes trial-and-error using a small batch of design configurations, which may not lead to an optimal solution. Usually, the heavy reliance on guesswork and the resulting subjectivity lead to inaccurate assumptions.

3.3 Prototyping Method #3: CAD, DHM, and Surrogate Models.

Prototyping Method #3 uses surrogate models, in addition to CAD and DHM, to represent and explore the design space. The surrogate modeling is an approximation method that is used for evaluating design objectives and constraint functions when real models are not available. This approach has been used in many engineering studies as a computationally cheaper methodology to explore design spaces when an outcome of interest cannot be directly measured [6166]. The surrogate modeling technique presented in Method #3 is adapted from a previous study of Ahmed et al. [67]. The study uses a Kriging modeling technique to enable designers to tie human performance to the design variables by systematically assessing human performance for a large number of design configurations. In the surrogate modeling approach, designers first change one design variable at a time to observe the variation in human performance outcomes and use statistics to identify the design variables that significantly affect human performance. Once the design variables are identified, the Latin hypercube sampling (LHS) method [68] is used to generate sample design configurations. Next, human performance data for each design configuration are extracted using DHM to create a Kriging surrogate model [62]. The surrogate model is then explored to find the design configuration that gives optimal human performance. Since the surrogate modeling approach has a higher fidelity when compared to prototyping Methods #1 and #2, it reduces the designer’s subjectivity by enabling a more systematic design space exploration.

4 Case Study

In this paper, three computational prototyping methodologies with different fidelity levels (low-, medium-, and high-fidelity) are compared to study their adequacy for evaluating ergonomics of products that comprise low to high levels of human–product interactions. Method #1, Method #2, and Method #3 are used as prototyping strategies to perform computational ergonomics analyses on a generic wall-mounted cabinet, an automobile steering wheel, an assembly line, and a simplified cockpit model. These case studies contain varying levels (low and high levels) of human–product interactions, which require different types of ergonomic analyses, as shown in Fig. 2. Design variables, design objectives, and types of ergonomic assessments are listed at the bottom of Fig. 2 for each case study. The level of human–product interaction increases from left to right (from cabinet to cockpit #2), which is also evident by the increase in the number of variables and objectives. The vertical axis, which ranges from low to high, represents prototyping fidelity levels. Prototyping Method #1 has the lowest fidelity among the three as it only has a sketching tool without any embedded ergonomic analysis capability, as shown in Fig. 2. Prototyping Method #2 has higher fidelity than Method #1 because, in this approach, CAD is used to represent the workplace and DHM is used for performing ergonomic analysis, as shown in Fig. 2. Prototyping Method #3 has the highest fidelity among the three because not only it uses CAD and DHM but, additionally, it implements surrogate modeling and optimization to explore larger design space and generate a larger solution space. Prototyping Method #3 uses the LHS method to generate multiple configurations of the workplace, as shown in the top row of the Method #3 in Fig. 2. The ergonomic analysis for each configuration is performed, and the generated data are used to create the surrogate model, as shown in the second and third rows, respectively. Finally, optimization is used to explore the design space and find the design configuration where the human performance is maximum, as shown in the fourth row. Compared to Method #2, these additional steps of LHS, surrogate modeling, and optimization in Method #3 reduce the designer bias when generating workplace configurations and design space exploration, thus increases the fidelity. Note that the illustrations for LHS in Fig. 2, shown in the Method #3 row on the y-axis, represents generic surrogate modeling models not specific optimization results for each study.

In this study, a 5th percentile Japanese female anthropometry is considered as the computational manikin model to represent the near-smallest population percentile in ergonomic assessments. It is because many conventional consumer products and workplace designs focus on the “average” users and ignore the population extremities. Often, a majority of the ergonomics issues regarding accessibility are associated with users from anthropometric population extremities. The design objectives and variables used in the study are provided in Fig. 2. The ranges (e.g., maximum and minimum reach envelop measures for the manikin) of the design variables are gathered according to the anthropometric extremities of the manikins used in this study and the consumer databases corresponding to the products [69].

4.1 Low-Level Interaction: Cabinet.

In this study, the cabinet model represents a generic product that has low-level human–product interaction. The scenario considered is someone trying to reach a specific point in the cabinet. Ideally, a cabinet needs to be designed with sufficient space so that it can hold as many items as possible and, at the same time, allows users ease of access. The cabinet geometry represents a simple form factor (Fig. 3, CAD model). The height and length are the design variables, and reachability is the only ergonomic factor affecting human interaction. As a result, the cabinet has a simple design space with low-level human–product interactions, which requires a relatively simple ergonomic assessment. Thus, in this study, the first objective is to increase reachability. It involves minimizing the reach gap between the manikin’s index fingertip and the cabinet corners, enabling the manikin to access all four inner corners of the cabinet. And the second objective is to maximize the cabinet area.

4.2 Low-Level Interaction: Automobile Steering Wheel.

This case study’s design objective is to provide maximum vision coverage (forward binocular vision) to the driver since the steering wheel location can affect the visibility of the dashboard and road elements (e.g., road signs, pedestrians, other vehicles) which negatively affect the driver’s performance by increasing obscuration zones. Design variables of interest are the vertical position and tilt angle of the steering wheel. This scenario presents a low-level human–product interaction example, as there is only one design objective (maximum vision coverage) and two design variables (vertical position and tilt angle) to consider. The ergonomic assessment required for this study is vision coverage.

4.3 Low- to Mid-Level Interaction: Assembly Lines #1 and #2.

Assembly Line #1 has similar ergonomics requirements and interaction levels as the cabinet study. The manikin is expected to reach each corner of the assembly line without bending. The design objective is to minimize the reach gap and maximize the surface area. Assembly Line #2, on the other hand, has a different level of human–product interaction when compared to Assembly Line #1. Thus, the study requires additional ergonomic assessments. In the Assembly Line #2 scenario, the manikin is allowed to bend forward while reaching the corners of the Assembly Line. Therefore, the manikins’ L4/L5 (compression force measurements between 4th and 5th lumbar sections) need to be evaluated for ergonomic adequacy. As a result, the design objectives are increased from two to three, with the addition of the L4/L5 measurements.

4.4 High-Level Interaction: Boeing Cockpits #1 and #2.

The aviation sector uses HFE guidelines heavily to evaluate pilot and crew performance. Within a cockpit environment, pilots interact with several objects such as instrument panels, yoke, pedestal, displays, and controls. Also, pilots need to have unobscured visual detection through the windshield, especially during take-off and landing. Two cockpit models are used to represent high-level human interaction scenarios. Although both studies share the same cockpit environment and have the same number of design variables (Fig. 4), there are different types of ergonomic assessments executed. For the Cockpit #1 case study, only the pilot’s reach task is assessed, whereas in Cockpit #2, both reach and vision tasks are evaluated. Hence, in Cockpit #1, the objective is to minimize the reach gap, whereas in Cockpit #2, the aim is to reduce the reach gap while maximizing the vision coverage.

5 Results

Six concept models were developed using the three computational prototyping methodologies (Methods #1, #2, and #3). The relevant ergonomic assessments (Fig. 2) were performed on each prototype to explore whether the differences in fidelity and the human–product interaction levels have any effects on the quality of the ergonomics outcomes. The results are discussed in this section.

5.1 Prototyping Method #1: Sketchpad.

Figure 2 shows the sketches representing a 5th percentile Japanese female interacting with all six products. The manikin sketch in Method #1 does not accurately represent population percentiles (e.g., 5th percentile Japanese female in this case), since the sketchpad tool does not contain any integrated anthropometric data. Therefore, each manikin had to be sketched manually based on the Japanese 2006 anthropometric database as stick-figures. The length of a 5th percentile Japanese female arm, from acromioclavicular joint (joint at the top of the shoulder) to index fingertip, was found to be around 62.53 cm. This information is used to manually sketch a circle with an approximate radius of 62.53 cm to illustrate a representative reach envelope (2D semi-circle). It should be noted that only the reach assessment can be evaluated using this approach. Ergonomic assessments of the vision coverage and L4/L5 analysis cannot be executed using a 2D sketchpad, which is a limitation of the Method #1.

The wall-mounted cabinet, as shown in Fig. 2, was kept at a constant shoulder height of 121.61 cm above the ground and 15 cm away from the manikin. Since the cabinet dimensions are symmetrical and the left arm was solely used during the reaching task, only one half of the cabinet was utilized during the ergonomics assessment. The reach envelope, as shown in Fig. 2, was assumed to be a sphere. The geometrical relations, such as the largest rectangle that can be inscribed in a circle, were used to explore the cabinet’s length and height. Since a square has the largest area inside a circle, the length and height of the cabinet are found to be around 44.21 cm on each side, resulting in a cabinet configuration that has the largest area within the reach envelope. The results of the reach assessment were shown in Table 1.

The sketchpad tool in Method #1 was successfully used to assess the reach envelope for products with a low number of design variables and objectives (see Table 1). However, when the number of design variables are high, Method #1 is not capable of replicating ergonomics assessments for reach envelope analysis. For example, the Cockpit #1 study has only one ergonomics design objective (reach envelope) similar to the Cabinet case study. However, the ergonomics evaluation cannot be performed for Cockpit #1 via the sketchpad because there are five design variables, making it infeasible to represent within a 2D space. Likewise, it was not possible to assess the obscuration zones as well as the lower-back analysis using the Sketchpad approach.

5.2 Prototyping Method #2: CAD and DHM.

A CAD file for each product was created and exported into DHM software (Siemens jack) for running ergonomic assessments. Unlike prototyping Method #1, the Method #2 can assess more complex ergonomics analyses that require 3D posture evaluation and interaction with the CAD environment, such as L4/L5, vision coverage, and reach assessment. A 5th percentile Japanese female manikin was created to perform a reach assessment for the cabinet, assembly line, and cockpit case studies. A reach envelope that has the shape of a bubble/sphere was generated in jack by tracing the tip of the left arm index finger of the manikin (Fig. 2). The translucent bubble represents the volume in which the manikin has extended reach when using the left arm. In the next step, various configurations of the cabinet, assembly line, and cockpit models were created by changing design variables. Each configuration includes the points of interest that were within the reach envelope. This approach ensures that the manikin has the reach coverage for all points of interest for each product. The results are presented in Table 1. For instance, the length and height of the cabinet were found to be around 45 cm. The process was iterated multiple times to achieve better consistency.

For the steering wheel study, the design objective was to maximize the drivers’ vision of the dashboard and windshield. The binocular vision coverage assessment tool of jack was used to evaluate the vision coverage. The design variables were the height and angle of the steering wheel. As the steering wheel was manually moved up and down and tilted, the optimal vision coverage percentage values were measured, as shown in Table 1. The L4/L5 analysis for the Assembly Line #2 case study was performed using a similar approach, where the design variables were manually adjusted to find a configuration with the minimum L4/L5 forces.

It should be noted that the cabinet, steering wheel, Assembly Line #1, and Cockpit #1 case studies require only one ergonomic assessment, whereas the Assembly Line #2 and Cockpit #2 require multiple ergonomic assessments. When only one ergonomic evaluation is desired, designers can focus solely on that ergonomic assessment and find the optimal design solution without running exhaustive design space exploration. However, when multiple ergonomic assessments are present, designers need to carefully configure the design variables and search the design space such that human performance is optimal for all ergonomic measures that are under evaluation.

In the Cockpit #2 study, the design objectives were to minimize the reach gap and maximize the vision coverage. Figure 2 shows the reach and vision coverage assessments for a 5th percentile Japanese female manikin placed inside the cockpit model. The values of the five design variables are changed according to the designer’s subjective choice when creating new cockpit configurations, which were then evaluated for instrument panel reach and vision coverage assessment. Reach assessment was calculated by measuring the reach gap between the left-hand index finger of the manikin and the surface of the instrument panel. Vision analysis was performed by calculating the percentage of the visible area of the windshields. In Fig. 2, the dotted lines show the visible region and the obscured region. It should be noted that identifying the optimal configuration that results in minimum reach gap and maximum vision coverage, it is up to the designer’s expertise and knowledge and may take multiple iterations. These back-and-forth iterations are repeated many times until the designer is satisfied with the design solution. The reach gap and vision coverage results are shown in Table 1.

5.3 Prototyping Method #3: CAD, DHM, and Surrogate.

As mentioned in Sec. 3.3, Method #3 applies surrogate modeling along with CAD and DHM [67,70]. Therefore, Method #3 shares the same prototyping environment as Method #2 with the addition of surrogate model-based optimization study (see Fig. 2). Although Method #2 and Method #3 have identical ergonomics assessments (reach gap, L4/L5, and vision coverage), the approach for creating new product configurations is different. In prototyping Method #3, product configurations are created using LHS, which eliminates the subjectivity in creating new configurations as observed in prototyping Method #2. LHS enables designers to generate configurations that cover a broad spectrum of design variables, representing a larger design space. Furthermore, by running optimization using the surrogate model, the best product or workplace configurations that can lead to optimal human performance can be identified.

For prototyping Method #3, the design objectives are to create products that have a minimum reach gap, minimum L4/L5, and maximum vision coverage. For example, for the cabinet, a square shape with a 45.6 cm length is found to have the largest coverage area with a minimum reach gap. Unlike the cabinet, the assembly line and cockpit case studies have multiple design objectives that are contradictory to each other. Pareto fronts are built for those products that have conflicting design objectives. Pareto fronts help in visualizing the tradeoffs between the two design objectives and selecting design configurations based on the design objective priorities. For example, the Pareto front of the Cockpit #2 case study is shown in Fig. 5, where one design objective gets better while the other gets worse.

To improve consistency, the optimization of the surrogate model is iterated multiple times. The results of the ergonomic assessments performed using this approach are available in Table 1. Overall, Table 1 summarizes the results of the ergonomic assessments performed using each prototyping method (Methods #1, #2, and #3). Note that the cockpit model could not be designed accurately using the sketchpad, so no result is shown in Table 1. The results of the cockpit model using prototyping Methods #2 and #3 are analyzed further. An independent two-sample t-test is used to identify whether designing the cockpit using different prototype fidelity yields any differences in human performance. The t-test results are shown in Table 2. The p-values indicate that the results obtained from the two prototyping strategies are significantly different from each other.

6 Discussion

This research aims to study the intricacies between the fidelity level, human–product interaction level, and ergonomic assessment and how it can be adopted in the Industry 4.0 paradigm to enable computational prototyping to be part of the product development. Successful human-centered prototyping strategies can help in mass customization by minimizing the product development cost and time. Hence, in this paper, we study the level of fidelity needed in computational prototypes when performing different types of ergonomic assessments on products with varying levels of human–product interactions. Six concept products with different ergonomic assessment requirements (reach, vision, and L4/L5) and different levels of human–product interactions (low, mid, and high) are prototyped using (1) prototyping Method #1 (low-fidelity): digital sketchpad, (2) prototyping Method #2 (mid-fidelity): CAD and DHM, and (3) prototyping Method #3 (high-fidelity): CAD, DHM, and surrogate modeling. The results and statistical analysis are presented in Tables 1 and 2. Several noteworthy prototyping findings extracted from this study are summarized in the following paragraphs.

Table 1 shows that the low-fidelity prototyping tool is suitable for evaluating ergonomics for the cabinet and Assembly Line #1 case studies. One can see that the reach assessment can be performed with the low-fidelity sketchpad since it involves only simple geometry calculations. As seen in Table 1, the reach gap measurements for the cabinet and Assembly Line #1 were found to be zero, using all three prototyping methods. Thus, all three methods are suitable for performing the reach assessment for products that contain low- to mid-level human interactions. On the other hand, prototyping Method #1 was not useful when assessing the reach in the Cockpit #1 scenario. Even though the Cockpit #1 model only required just the reach assessment, performing reach analysis with a 2D sketchpad was infeasible. Five design variables in total and a high level of human interaction make the low-fidelity prototyping tool sketchpad not suitable for this case study. Thus, one can see that Method #1 is not an appropriate approach to deliver the reach assessments for products that contain high levels of human interaction. When a design problem includes a high-level human–product interaction and 3D configurations, using a low-fidelity computational prototyping method, as seen in Method #1, is not recommended.

Results in Table 1 show that the mid- and high-fidelity level prototyping methods were suitable for prototyping all six concept products. The question here is whether the difference in fidelity levels affect the design solutions. The answer lies in the results of the independent two-sample t-test in Table 2. A random sample is selected and the normal distribution of the data is checked before the t-test is carried out. All p-values are significant (p < 0.05) except for reach gap in Assembly Line #2, meaning that there was a difference in the ergonomic assessment results between the prototyping strategies. The difference in the design solutions can be attributed to the fidelity level of the prototype. As mentioned before, designers use their expertise when working with prototyping Method #2, which often leads to an under-exploration of the design space. Whereas, in prototyping with Method #3, a surrogate model is built to find optimal design space using optimization techniques. Prototyping Method #3 eliminates a majority of the subjectivity and aid designers in exploring the design space systematically to find better solutions as compared to prototyping Method #2. The differences between the capability of prototyping Method #2 and prototyping Method #3 become more apparent as the design space or human–product interaction increases. For example, the p-value for the reach gap of Assembly Line #2 is close to 0.05 (i.e., the mean value obtained from prototyping Method #2 and prototyping Method #3 is not different). This indicates that the design space related to the reach gap assessment is small when there is a low-level interaction or only two design variables to consider.

Additionally, it is also observed that Methods #2 and #3 produce contrasting outcomes when both design objectives are equally prioritized. One can see that combined mean differences are higher and p-values are lower. It is because, in prototyping Method #2, designers only manipulate the design variables in a limited number of ways to create the cockpit design configurations, which can achieve both design objectives equally. However, in prototyping Method #3, these limitations are eliminated with the use of the surrogate model and optimization tool.

Table 1 shows that all three prototypes are suitable for prototyping products that include the reach assessment and low-level human–product interactions. Moreover, Table 2 presents that higher fidelity prototypes are better in prototyping products that possess higher levels of human–product interactions. This raises a question: if all levels of fidelity are appropriate to use for product design, then what level of fidelity should one choose? To address this question, one needs to consider other factors, such as the cost of resources that go into building a prototype. Using the cabinet scenario as an example, prototyping using a sketchpad takes around 3–5 min, finding the approximate anthropometry data and dimensions takes another 3–5 min. Finally, the calculations take around another minute or two. Therefore, prototyping using a sketchpad option takes approximately 10 min. In contrast, prototyping Method #2 takes about 15–20 min as the designer needs to create various CAD files and test ergonomics using DHM software. The prototyping Method #3 takes the longest time of approximately an hour. Prototyping Method #3 requires creating configurations of design variables, creating a surrogate model, and using an optimization tool to explore the design space. Also, prototyping Methods #2 and #3 are costlier than prototyping Method #1 due to the software expenses and time commitments (e.g., jack, SolidWorks, and matlab). As the level of fidelity increases (from Method #1 to #3), the cost of resources in terms of time and money also increases; however, the cabinet configurations and ergonomic assessment yield results that are close to each other. Hence, a low-fidelity computational prototype can be suitable to design a product or workplace with low-level human interactions and a limited design configurations (small design space).

To sum up, one can conclude that the low-fidelity prototyping approach (prototyping via a 2D sketchpad tool) is limited in products with high human–product interactions. Also, low-fidelity prototyping has limitations in performing complex ergonomic assessments, for example, for the vision and L4/L5 analyses. Yet, the low-fidelity prototyping approach can still be used in executing ergonomics studies in scenarios that include low-level human–product interactions. If the human–product interaction level is not high, then low-fidelity prototyping produces outcomes comparable to mid- and high-level fidelity prototypes while using fewer resources. Mid- and high-level fidelity prototypes are recommended when high-level human–product interactions are available and when designers are interested in applying a wide range of ergonomic assessments. However, as the level of human–product interaction increases, the difference in the accuracy between the mid-fidelity and high-fidelity prototype results becomes more prominent. A high-fidelity prototype produces more accurate results than a mid-level prototype because it enables designers to do a more thorough and objective design space search. Overall, it can be suggested that designers must decide which fidelity level prototype to use after doing a tradeoff study between the accuracy of the ergonomics outcomes and the available resources.

7 Conclusion and Future Work

Although the Industry 4.0 concept promises better control of the overall product development process, there is a lack of computational frameworks that can inject ergonomics and HFE early in design. This shortage is particularly crucial for prototyping human-centered products where the stakes are high. Ergonomically inferior products are associated with reduced quality, lousy safety records, and low levels of user satisfaction. Overall, they accumulate an extensive cost to manufacturers in the long run due to product recalls, loss in market share, and diminished customer loyalty. Thus, a smooth transformation within the Industry 4.0 paradigm cannot occur without bringing computational ergonomics tools and methodologies into the loop, specifically to address the needs in the digitized prototyping process. In this study, a unique computational ergonomics approach is demonstrated to solve some of the above shortcomings.

One of the limitations of this study is that it only focused on computational prototyping approaches. The prototyping findings may apply to other types of prototypes, such as physical and mixed prototypes, but require further studies for validation. Also, the results presented in this research are generated through computational models and not validated within an actual product development practice. The results can be validated by replacing the computational prototypes with low-, mid-, and high-fidelity physical prototypes and substituting digital manikins with the actual users for human subject ergonomics data collection. Another limitation of this study is that only three types of ergonomic assessments (reach, vision, and L4/L5) are considered, making the results only valid for a limited scope of ergonomics evaluations.

One avenue of future work is to develop a computational prototyping framework that integrates prototyping best practices with ergonomics and human factor guidelines to guide designers and engineers to prototype human-centered products. Currently, no prototyping framework considers ergonomics and human factors guidelines concurrently. However, the current research findings related to prototyping that considers ergonomics are inadequate. Hence, to develop the prototyping framework for a human-centered product, research studies related to other ergonomics assessments such as fatigue, strength analysis, comfort analysis, and cognitive analysis need to be performed initially. We expect these research studies to provide design guidelines and best practices regarding human-centered prototyping activities, leading to the creation of more comprehensive computational prototyping frameworks to support Industry 4.0 objectives. Furthermore, the computational human-centered product prototyping framework can be integrated with a graphical user interface or with next-generation computer-aided engineering tools to automate the prototyping process and generate conceptual prototyping strategies. The automation and digitization of the prototyping process would fit into the overarching goal of Industry 4.0. of achieving higher efficiency and productivity.

Conflict of Interest

There are no conflicts of interest.

Data Availability Statement

The authors attest that all data for this study are included in the paper.

References

1.
Schlaepfer
,
R.
,
Koch
,
M.
, and
Merkofer
,
P.
,
2015
, “
Challenges and Solutions for the Digital Transformation and Use of Exponential Technologies
,” Deloitte, https://www2.deloitte.com/content/dam/Deloitte/ch/Documents/manufacturing/ch-en-manufacturing-industry-4-0-24102014.pdf
2.
Lin
,
S.
,
Mellor
,
S.
,
Miller
,
B.
,
Durand
,
J.
,
Crawford
,
M.
, and
Lembree
,
R.
,
2015
, “
Industrial Internet Reference Architecture, Version 1.7
,” Industrial Internet Consortium, https://www.iiconsortium.org/IIRA-1-7-ajs.pdf, Zugegriffen am,
3
, p.
2020
.
3.
The Industry of the Future—France in the United Kingdom—la france au royaume-uni
,”
2020
, https://uk.ambafrance.org/The-Industry-of-the-Future, Accessed August 29, 2020.
4.
‘Made in China 2025’ Industrial Policies: Issues for Congress
,”
2020
, https://fas.org/sgp/crs/row/IF10964.pdf, Accessed August 29, 2020.
5.
Rojko
,
A.
,
2017
, “
Industry 4.0 Concept: Background and Overview
,”
Int. J. Int. Mobile Technol. (iJIM)
,
11
(
5
), pp.
77
90
.
6.
Bartodziej
,
C. J.
,
2017
, “The Concept Industry 4.0,”
The Concept Industry 4.0
,
Springer
, pp.
27
50
.
7.
Almada-Lobo
,
F.
,
2015
, “
The Industry 4.0 Revolution and the Future of Manufacturing Execution Systems (MES)
,”
J. Innov. Manage.
,
3
(
4
), pp.
16
21
.
8.
Schlechtendahl
,
J.
,
Keinert
,
M.
,
Kretschmer
,
F.
,
Lechler
,
A.
, and
Verl
,
A.
,
2015
, “
Making Existing Production Systems Industry 4.0-Ready
,”
Prod. Eng.
,
9
(
1
), pp.
143
148
.
9.
Roblek
,
V.
,
Meško
,
M.
, and
Krapež
,
A.
,
2016
, “
A Complex View of Industry 4.0
,”
Sage Open
,
6
(
2
), p.
2158244016653987
.
10.
Babiceanu
,
R. F.
, and
Seker
,
R.
,
2016
, “
Big Data and Virtualization for Manufacturing Cyber-physical Systems: A Survey of the Current Status and Future Outlook
,”
Comput. Ind.
,
81
, pp.
128
137
.
11.
Dalenogare
,
L. S.
,
Benitez
,
G. B.
,
Ayala
,
N. F.
, and
Frank
,
A. G.
,
2018
, “
The Expected Contribution of Industry 4.0 Technologies for Industrial Performance
,”
Int. J. Prod. Econ.
,
204
, pp.
383
394
.
12.
Lee
,
J.
,
Bagheri
,
B.
, and
Kao
,
H.-A.
,
2015
, “
A Cyber-physical Systems Architecture for Industry 4.0-Based Manufacturing Systems
,”
Manuf. Lett.
,
3
, pp.
18
23
.
13.
Chen
,
Q.
,
Heydari
,
B.
, and
Moghaddam
,
M.
,
2021
, “
Levering Task Modularity in Reinforcement Learning for Adaptable Industry 4.0 Automation
,”
ASME J. Mech. Des.
,
143
(
7
), p.
071701
.
14.
Posada
,
J.
,
Toro
,
C.
,
Barandiaran
,
I.
,
Oyarzun
,
D.
,
Stricker
,
D.
,
De Amicis
,
R.
,
Pinto
,
E. B.
,
Eisert
,
P.
,
Döllner
,
J.
, and
Vallarino
,
I.
,
2015
, “
Visual Computing as a Key Enabling Technology for Industrie 4.0 and Industrial Internet
,”
IEEE Comput. Graph. Appl.
,
35
(
2
), pp.
26
40
.
15.
Wang
,
R.
,
Milisavljevic-Syed
,
J.
,
Guo
,
L.
,
Huang
,
Y.
, and
Wang
,
G.
,
2021
, “
Knowledge-Based Design Guidance System for Cloud-Based Decision Support in the Design of Complex Engineered Systems
,”
ASME J. Mech. Des.
,
143
(
7
), p.
072001
.
16.
Lu
,
Y.
,
2017
, “
Industry 4.0: A Survey on Technologies, Applications and Open Research Issues
,”
J. Ind. Inform. Int.
,
6
, pp.
1
10
.
17.
Ahmed
,
S.
,
Irshad
,
L.
,
Demirel
,
H. O.
, and
Tumer
,
I. Y.
,
2019
, “A Comparison Between Virtual Reality and Digital Human Modeling for Proactive Ergonomic Design,”
International Conference on Human–Computer Interaction
,
Duffy
,
V. G.
, ed.,
Lecture Notes in Computer Science: Digital Human Modeling and Applications in Health, Safety, Ergonomics and Risk Management. Human Body and Motion
,
Springer International Publishing
,
Orlando, FL
, pp.
3
21
.
18.
Frank
,
A. G.
,
Dalenogare
,
L. S.
, and
Ayala
,
N. F.
,
2019
, “
Industry 4.0 Technologies: Implementation Patterns in Manufacturing Companies
,”
Int. J. Prod. Econ.
,
210
, pp.
15
26
.
19.
Demirel
,
H. O.
,
2020
, “
Digital Human-in-the-Loop Framework
,”
International Conference on Human–Computer Interaction
,
Copenhagen, Denmark
,
July 19–24
,
Springer
, pp.
18
32
.
20.
Stanton
,
N. A.
, and
Young
,
M. S.
,
2003
, “
Giving Ergonomics Away? The Application of Ergonomics Methods by Novices
,”
Appl. Ergon.
,
34
(
5
), pp.
479
490
.
21.
Naumann
,
A.
, and
Rötting
,
M.
,
2007
, “
Digital Human Modeling for Design and Evaluation of Human–Machine Systems
,”
MMI-Int.
,
12
, pp.
27
35
.
22.
Ferrise
,
F.
,
Bordegoni
,
M.
, and
Cugini
,
U.
,
2013
, “
Interactive Virtual Prototypes for Testing the Interaction With New Products
,”
Comput. Aided Des. Appl.
,
10
(
3
), pp.
515
525
.
23.
Broek
,
J. J.
,
Sleijffers
,
W.
,
Horváth
,
I.
, and
Lennings
,
A. F.
,
2000
, “
Using Physical Models in Design
,”
Proceedings of CAID/CD’2000 Conference
, pp.
155
163
.
24.
Bi
,
Z.
,
2010
, “
Computer Integrated Reconfigurable Experimental Platform for Ergonomic Study of Vehicle Body Design
,”
Int. J. Comput. Int. Manuf.
,
23
(
11
), pp.
968
978
.
25.
Bullinger
,
H.-J.
, and
Dangelmaier
,
M.
,
2003
, “
Virtual Prototyping and Testing of In-vehicle Interfaces
,”
Ergonomics
,
46
(
1–3
), pp.
41
51
.
26.
Kuutti
,
K.
,
Battarbee
,
K.
,
Sade
,
S.
,
Mattelmaki
,
T.
,
Keinonen
,
T.
,
Teirikko
,
T.
, and
Tornberg
,
A.-M.
,
2001
, “
Virtual Prototypes in Usability Testing
,”
Proceedings of the 34th Annual Hawaii International Conference on System Sciences
,
Maui, HI
,
IEEE
, p.
7
.
27.
Duffy
,
V. G.
,
2007
, “
Modified Virtual Build Methodology for Computer-Aided Ergonomics and Safety
,”
Hum. Factors Ergon. Manuf. Serv. Ind.
,
17
(
5
), pp.
413
422
.
28.
Ahmed
,
S.
,
Irshad
,
L.
, and
Demirel
,
H. O.
,
2019
, “
Computational Prototyping Methods to Design Human Centered Products of High and Low Level Human Interactions
,”
International Design Engineering Technical Conferences and Computers and Information in Engineering Conference
,
Anaheim, CA
,
Aug. 18–21
, Vol.
59278
,
American Society of Mechanical Engineers
, p.
V007T06A047
.
29.
Otto
,
K. N.
and
Wood
,
K. L.
,
2003
,
Product Design: Techniques in Reverse Engineering and New Product Development
,
Prentice Hall
,
Englewood Cliffs, NJ
.
30.
Pahl
,
G.
, and
Beitz
,
W.
,
2013
,
Engineering Design: A Systematic Approach
,
Springer Science & Business Media
,
Berlin, Germany
.
31.
Zorriassatine
,
F.
,
Wykes
,
C.
,
Parkin
,
R.
, and
Gindy
,
N.
,
2003
, “
A Survey of Virtual Prototyping Techniques for Mechanical Product Development
,”
Proc. Inst. Mech. Eng. B
,
217
(
4
), pp.
513
530
.
32.
Stowe
,
D.
,
2008
, “
Investigating the Role of Prototyping in Mechanical Design Using Case Study Validation
,”
All Theses, 532
, https://tigerprints.clemson.edu/all_theses/532
33.
Jönsson
,
A.
,
2004
, “
Lean Prototyping of Multi-body and Mechatronic Systems
,”
PhD thesis
,
Blekinge Institute of Technology
,
Karlskrona
.
34.
Tseng
,
M. M.
,
Jiao
,
J.
, and
Su
,
C.-J.
,
1997
, “
A Framework of Virtual Design for Product Customization
,”
1997 IEEE 6th International Conference on Emerging Technologies and Factory Automation Proceedings, EFTA’97
,
Los Angeles, CA
,
Sept. 9–12
,
IEEE
, pp.
7
14
.
35.
Hansen
,
C. A.
, and
Özkil
,
A. G.
,
2020
, “
From Idea to Production: A Retrospective and Longitudinal Case Study of Prototypes and Prototyping Strategies
,”
ASME J. Mech. Des.
,
142
(
3
).
36.
Brereton
,
M.
, and
McGarry
,
B.
,
2000
, “
An Observational Study of How Objects Support Engineering Design Thinking and Communication: Implications for the Design of Tangible Media
,”
Proceedings of the SIGCHI Conference on Human Factors in Computing Systems
,
April
,
ACM
, pp.
217
224
.
37.
Alexopoulos
,
K.
,
Mavrikios
,
D.
, and
Chryssolouris
,
G.
,
2013
, “
Ergotoolkit: An Ergonomic Analysis Tool in a Virtual Manufacturing Environment
,”
Int. J. Comput. Int. Manuf.
,
26
(
5
), pp.
440
452
.
38.
Boothe
,
C.
,
Strawderman
,
L.
, and
Hosea
,
E.
,
2013
, “
The Effects of Prototype Medium on Usability Testing
,”
Appl. Ergon.
,
44
(
6
), pp.
1033
1038
.
39.
Walker
,
M.
,
Takayama
,
L.
, and
Landay
,
J. A.
,
2002
, “
High-Fidelity or Low-Fidelity, Paper or Computer? Choosing Attributes When Testing Web Prototypes
,”
Proceedings of the Human Factors and Ergonomics Society Annual Meeting
,
Sept. 1
, Vol.
46
,
SAGE Publications Sage
,
Los Angeles, CA
, pp.
661
665
.
40.
Hall
,
R. R.
,
2001
, “
Prototyping for Usability of New Technology
,”
Int. J. Hum. Comput. Stud.
,
55
(
4
), pp.
485
501
.
41.
Ward
,
S.
,
1994
, “
Getting Feedback From Users Early in the Design Process: A Case Study, Ergonomics: The Fundamental Design Science
,”
Proceedings of the 30th Annual Conference of the Ergonomics Society of Australia
,
The Ergonomics Society of Australia Canberra
, pp.
22
29
.
42.
Virzi
,
R. A.
,
Sokolov
,
J. L.
, and
Karis
,
D.
,
1996
, “
Usability Problem Identification Using Both Low- and High-Fidelity Prototypes
,”
Proceedings of the SIGCHI Conference on Human Factors in Computing Systems
,
Waltham, MA
,
ACM
, pp.
236
243
.
43.
Camburn
,
B.
,
Dunlap
,
B.
,
Gurjar
,
T.
,
Hamon
,
C.
,
Green
,
M.
,
Jensen
,
D.
,
Crawford
,
R.
,
Otto
,
K.
, and
Wood
,
K.
,
2015
, “
A Systematic Method for Design Prototyping
,”
ASME J. Mech. Des.
,
137
(
8
), p.
081102
.
44.
Chaffin
,
D. B.
,
2009
, “Some Requirements and Fundamental Issues in Digital Human Modeling,”
Handbook of Digital Human Modeling
,
V.
Duffy
, ed.,
CRC Press
,
Boca Raton, FL
, pp.
2
1
.
45.
Demirel
,
H. O.
, and
Duffy
,
V. G.
,
2007
, “
Applications of Digital Human Modeling in Industry
,”
International Conference on Digital Human Modeling
,
Beijing, China
,
July 22–27
,
Springer
, pp.
824
832
.
46.
van der Meulen
,
P.
, and
Seidl
,
A.
,
2007
, “
Ramsis—The Leading CAD Tool for Ergonomic Analysis of Vehicles
,”
International Conference on Digital Human Modeling
,
Beijing, China
,
July 22–27
,
Springer
, pp.
1008
1017
.
47.
Santos
,
J.
,
Sarriegi
,
J. M.
,
Serrano
,
N.
, and
Torres
,
J. M.
,
2007
, “
Using Ergonomic Software in Non-repetitive Manufacturing Processes: A Case Study
,”
Int. J. Ind. Ergon.
,
37
(
3
), pp.
267
275
.
48.
Wang
,
W.
,
Sun
,
Y. C.
, and
Kong
,
F.
,
2008
, “Analysis and Study of Ergonomics on Virtual Maintenance of Civil Airplane Based on Delmia,”
Advanced Materials Research
, Vol.
44
,
Trans Tech Publ.
, pp.
821
828
.
49.
Blanchonette
,
P.
,
2010
, “
Jack Human Modelling Tool: A Review
,”
Technical Report, Defence Science and Technology Organisation Victoria (Australia)
.
50.
Polášek
,
P.
,
Bureš
,
M.
, and
Šimon
,
M.
,
2015
, “
Comparison of Digital Tools for Ergonomics in Practice
,”
Procedia Eng.
,
100
, pp.
1277
1285
.
51.
Colombo
,
G.
,
Ponti
,
D. G.
, and
Rizzi
,
C.
,
2009
, “
Digital Human Modeling for Ergonomic Analysis of Refrigerated Cabinets
,”
ICORD 09: Proceedings of the 2nd International Conference on Research Into Design
,
Bangalore, India
,
Jan. 7–9
, pp.
331
338
.
52.
Wang
,
G. G.
,
2002
, “
Definition and Review of Virtual Prototyping
,”
ASME J. Comput. Inf. Sci. Eng.
,
2
(
3
), pp.
232
236
.
53.
Ianni
,
J.
,
2001
, “Human Model Evaluations of Air Force System Designs,”
Digital Human Modeling for Vehicle and Workplace Design
,
D. B.
Chaffin
, ed.,
SAE International
, pp.
37
55
.
54.
Ackerman
,
M. J.
,
2016
, “
The Visible Human Project(r): From Body to Bits
,”
Conf. Proc. IEEE Eng. Med. Biol. Soc.
,
2016
, pp.
3338
3341
.
55.
Green
,
R. F.
,
Hagale
,
T. J.
,
George
,
T.
,
Hancock
,
G. A.
, and
Rice
,
S. M.
,
2019
, “Digital Human Modeling in Aerospace,”
DHM and Posturography
,
S.
Scataglini
, and
G.
Paul
, eds.,
Elsevier
,
Chicago, IL
, pp.
549
558
.
56.
Choi
,
H. Y.
,
2010
, “
Potential Use of Digital Human Models in Sports Science and Technology
,”
J. Soc. Mech. Eng.
,
113
(
1095
), pp.
96
99
.
57.
Caputo
,
F.
,
Greco
,
A.
,
Fera
,
M.
, and
Macchiaroli
,
R.
,
2019
, “
Workplace Design Ergonomic Validation Based on Multiple Human Factors Assessment Methods and Simulation
,”
Prod. Manuf. Res.
,
7
(
1
), pp.
195
222
.
58.
Demirel
,
H. O.
,
2015
, “
Modular Human-in-the-Loop Design Framework Based on Human Factors
,”
PhD thesis
,
Purdue University
,
West Lafayette, IN
.
59.
Ahmed
,
S.
,
Zhang
,
J.
, and
Demirel
,
O.
,
2018
, “
Assessment of Types of Prototyping in Human-Centered Product Design
,”
International Conference on Digital Human Modeling and Applications in Health, Safety, Ergonomics and Risk Management
,
Las Vegas, NV
,
July 15–20
,
Springer
, pp.
3
18
.
60.
Sketchpad 5.1—Draw, Create, Share!
,”
2019
, https://sketch.io/sketchpad/, Accessed February 23, 2019.
61.
Lophaven
,
S. N.
,
Nielsen
,
H. B.
,
Sondergaard
,
J.
, and
Dace
,
A.
,
2002
, “
A matlab Kriging Toolbox
,”
Technical University of Denmark, Kongens Lyngby, Technical Report No. IMM-TR-2002-12
.
62.
Huang
,
Z.
,
Wang
,
C.
,
Chen
,
J.
, and
Tian
,
H.
,
2011
, “
Optimal Design of Aeroengine Turbine Disc Based on Kriging Surrogate Models
,”
Comput. Struct.
,
89
(
1–2
), pp.
27
37
.
63.
Ahmed
,
S.
,
Goh
,
C.-H.
,
Allen
,
J. K.
,
Mistree
,
F.
,
Zagade
,
P.
, and
Gautham
,
B.
,
2014
, “
Hot Forging of Automobile Steel Gear Blanks: An Exploration of the Solution Space
,”
ASME 2014 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference
,
Buffalo, NY
,
Aug. 17–20
,
American Society of Mechanical Engineers
, p.
V02BT03A003
.
64.
Ahmed
,
S.
,
Goh
,
C.-H.
,
Gautham
,
B. P.
,
Zagade
,
P.
,
Allen
,
J. K.
, and
Mistree
,
F.
,
2014
, “
Design Space Exploration With the Preemptive and Archimedean Formulations of the Compromise Decision Support Problem
,”
Int. J. Adv. Manuf. Sys.
,
15
(
1
), pp.
41
63
.
65.
Gorissen
,
D.
,
Couckuyt
,
I.
,
Demeester
,
P.
,
Dhaene
,
T.
, and
Crombecq
,
K.
,
2010
, “
A Surrogate Modeling and Adaptive Sampling Toolbox for Computer Based Design
,”
J. Mach. Learn. Res.
,
11
, pp.
2051
2055
.
66.
Audet
,
C.
,
Denni
,
J.
,
Moore
,
D.
,
Booker
,
A.
, and
Frank
,
P.
,
2000
, “
A Surrogate-Model-Based Method for Constrained Optimization
,”
8th Symposium on Multidisciplinary Analysis and Optimization
,
Long Beach, CA
,
Sept. 6–8
, p.
4891
.
67.
Ahmed
,
S.
,
Gawand
,
M. S.
,
Irshad
,
L.
, and
Demirel
,
H. O.
,
2018
, “
Exploring the Design Space Using a Surrogate Model Approach With Digital Human Modeling Simulations
,”
ASME 2018 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference
,
Quebec City, Quebec, Canada
,
Aug. 26–29
,
American Society of Mechanical Engineers
, p.
V01BT02A011
.
68.
McKay
,
M. D.
,
Beckman
,
R. J.
, and
Conover
,
W. J.
,
2000
, “
A Comparison of Three Methods for Selecting Values of Input Variables in the Analysis of Output From a Computer Code
,”
Technometrics
,
42
(
1
), pp.
55
61
.
69.
Goossens
,
R.
,
Snijders
,
C.
, and
Fransen
,
T.
,
2000
, “
Biomechanical Analysis of the Dimensions of Pilot Seats in Civil Aircraft
,”
Appl. Ergon.
,
31
(
1
), pp.
9
14
.
70.
Ahmed
,
S.
,
Irshad
,
L.
,
Gawand
,
M. S.
, and
Demirel
,
H. O.
,
2021
, “
Integrating Human Factors Early in the Design Process Using Digital Human Modelling and Surrogate Modelling
,”
J. Eng. Des.
,
32
(
4
), pp.
165
186
.