A New Configuration Method for Glass Substrate Transfer Robot Modules Based on Kansei Engineering

Featured Application

In the context of the widespread use of industrial robot modules, this study proposes a method for designing and evaluating the morphological characteristics and combination schemes of industrial robot modules, which meets the research and application needs of Kansei Engineering in robot module design.

Abstract

How to effectively match the relationship between users’ perceptual demands and the characteristics of industrial robot modules becomes a pressing issue when perceptual demands become a significant determinant of whether users purchase and employ industrial robots. In this regard, we propose a Kansei Engineering-based method for industrial robot module configuration, using the module design of a glass substrate transfer robot as an example. First, the method analyzes the perceptual demand characteristics of the target user, utilizing the semantic difference method, and then establishes a mapping relationship between the user’s perceptual demand and the robot design elements, utilizing the hierarchical inference method. On the basis of this mapping relationship, the robot module for transfer glass substrates is then designed. Finally, orthogonal design and conjoint analysis were used to effectively and objectively analyze user preferences for various module configuration alternatives. The results indicate that the industrial robot’s shape, color, and material are the three appearance characteristics that influence the user’s perceptual demands. The slender, rigid design features of the industrial robot, such as the streamlined drive shaft, lengthwise expanded body structure, integrated body structure, and hidden plugs, as well as the simple color scheme and smooth metal surface, are key elements in the industrial robot’s perceptual design. The turn shaft module and lift shaft module have respective weights of 35.040% and 31.120%, determining whether the glass substrate transfer robot can create a simple style. In the context of the widespread use of industrial robot modules, the methods and findings of this study offer new ideas for the design of industrial robot modules and broaden the research and applications of Kansei Engineering in module design.

1. Introduction

With the widespread adoption of modular design techniques for industrial robots, perceptual design has become crucial for industrial robots to gain a competitive advantage. To be specific, in the context of manufacturing industries’ increasing demand for industrial robots for large-scale production [1], industrial robots have been designed with a modular architecture to accommodate variable working environments and production requirements, as well as to meet flexible production demands [2]. In contrast to the industrial design of ordinary consumer goods, modular industrial robots are commonly viewed as the design of mechanical structures, for which the notion of the primacy of mechanical functionality has long dominated the design of industrial robot modules [3]. This unipolar development, which focuses on solving "functional and technical" problems, is likely to lose market share or even be eliminated [3,4] due to the increasing maturity and widespread use of industrial robotics and the increase in human emotion. In this study, for example, the glass substrate transfer robot must not only meet the mechanical performance requirements but also have a modular configuration that is consistent with the brand image of the intended user. According to a comprehensive perspective, as the industrial robot market becomes more competitive and product homogeneity becomes more prominent, people have new perceptions and demands about robot functioning. In other words, when it is difficult to differentiate robots based on “functional utility” and “technology”, excellent robot design will become key to market competition, and the assertion that “beautiful design is also functional” has become a hot topic for research on industrial robot design and its modular design.

Despite this, sufficient research is still lacking on the perceptual design of industrial robot modules at present. According to the literature reviews and practical research, the modular design of industrial robots at this stage is primarily concerned with functional utility, such as addressing the structure [5,6], power [7], path [8], and communication [9] of industrial robot modules through technical means. In the few perceptual design studies that have been conducted, the focus has also been on the design of the overall industrial robot, rather than the perceptual design of the individual modules of the robot. In the actual application of industrial robots, this results in the disadvantage that “one module configuration equals one redesign, and a new style is created”. This randomized, non-uniform, and non-standardized module configuration style does not meet the user’s requirements when the perceptual experience becomes a key factor influencing users to purchase and employ industrial robots [9,10]. To solve this implementation problem, we examine the Kansei Engineering-based method of industrial robot module configuration, using the module design of a glass substrate transfer robot as an example. The purpose of the study is to determine the mapping relationship between user perceptual demands and module design elements and then to design an industrial robot module and select the module configuration scheme based on these findings. Compared to previous research, this study presents a perceptual design and assessment of various combinations of modular shapes and solutions, which is useful for broadening the application of Kanseil Engineering in the design of industrial robot modules and serves as a reference for the design of industrial robot modules of the same type.

The remainder of the paper is structured as follows: In Section 2, we review the application of Kansei Engineering in the field of industrial robotics, based on the Kansei Engineering concept and the research challenges, and propose the problem to be addressed in this paper: how to carry out effective perceptual design oriented to the configuration and combination scheme of industrial robot modules with variable morphology. In Section 3, we propose a Kansei Engineering-based industrial robot module configuration method to solve this problem. In Section 4, we report the results of the study to verify the method’s validity, using the module design of a glass substrate transfer robot as an example. In Section 5, we discuss the findings, strengths, and limitations of the research methodology. Finally, in Section 6, we present the conclusions.

2. Related Work

“Perceptual demands” are a critical component of marketing to consumers in manufacturing and information service industries, which is supported by human-centered perceptual engineering or emotional design [11]. It is widely accepted that perception demands are a combination of complex human cognitive experiences, including abstract, highly ambiguous subjective feelings that are expressed in vague language and contain uncertain information [12]. Further, it has been noted that users often make perceptual judgments based on different values as a result of the differences in their own experiences and knowledge levels [13,14]. This means that the critical challenge of perceptual design lies in the translation of abstract user perception demands into concrete design elements [15]. To explore and practice human perceptual demands, researchers have proposed using “perceptual engineering”, a method that aims to accurately describe sensory and emotional experiences that may be difficult to explain subjectively and logically [16]. The concept of Kansei Engineering combines user research and engineering design, and related theories include the Kansei Engineering theory proposed by Japanese scholars and the three-level theory of emotional design proposed by American scholar Donald Arthur Norman [17], which is widely used in Korea and Japan. A similar conclusion has been reached by European and American academics: “Kansei Engineering is a valuable bridge between design, manufacturing, and human emotions” [18].

As technology advances and user demands change, Kansei Engineering has also gained application in industrial robot manufacturing. It is applied in two primary ways: creative design by incorporating user perceptions and emotions, and normative design by clarifying constraints and matching perceptual demands with design elements.

2.1. Identifying Industrial Robots’ Perceptual Demands and Emotional Experiences

Obtaining and configuring the user’s perceptual demands framework in a scientific and efficient manner is the primary challenge associated with the perception design of industrial robots. As a result of the complexity, abstraction, and multiplicity of a user’s perceptual demands, it is challenging for industrial designers to incorporate these demands into product design. To achieve effective perceptual demands, qualitative commentary vocabulary analysis and quantitative physiological index measurements are used, among other methods. Firstly, qualitative research is commonly used to identify the types of demands and affective attitudes of users in terms of the scope of application. As an example, some researchers have used personal interviews and questionnaires to extensively collect information on users’ demands for industrial robots and used the KJ classification method to perform hierarchical clustering and simplification of vulgar, similar, and opposite vocabularies. Based on these classified and generalized perceptual experiences and claims, the researcher identified the content of the perceptual demands of industrial robots [19]. Furthermore, some studies indicate that, in addition to qualitative vocabulary analysis, physiological indicators can be used to correlate the true emotional state of users in specific scenarios [20]. In this regard, some researchers have proposed using objective emotion measurement techniques, such as eye tracking, skin conductance, heart-rate variability, and brain electrophysiological activity (EEG), as well as combining subjective psychometric techniques, such as the semantic differential method and emotion scales, to determine the perception of industrial robots [21,22]. Finally, there are studies that use both qualitative and quantitative methods in the development of perceptual designs. Specifically, some researchers have found quantitative relationships between perceptual demands and total satisfaction using selection experiments of perceptual images, eye-movement experiments, and Kano models, and they developed small assistant robots accordingly [23].

2.2. Matching Industrial Robots’ Perceptual Demands and Design Elements

The practice of designing industrial robots according to the designer’s perceptions and creative ideas is always limited and may result in robots with unqualified mechanical performance [21]. Some studies have attempted to address this problem by analyzing the relationship between user perception demands and robot design elements (mechanical structures and components). As an example, some researchers have proposed that correlation analysis methods be used at the start of industrial robot design to assess the strength of the association between user-perceptual vocabulary and the product design elements of robots in order to help designers or engineers carry out an effective perceptual design process [24]. In a similar manner, to deal with the difficulty of mapping and expressing the emotions of the user, some researchers have developed a relationship matrix between “perceptual demands” and “design elements.” To construct this matrix, the researcher first used the rough analytic hierarchy process (RAHP) and quality function development (QFD) to determine the user’s perceptual design demands, followed by axiomatic design to determine the mapping relationship between perceptual design demands and robot components. Finally, the researchers used GRA-TOPSIS to measure the correlation strength between “perceptual demands” and “design elements” in order to select the robot design elements that best meet each user’s psychological and emotional demands [25].

By analyzing the existing research results, we made the following observations: First, in the field of industrial robot design, Kansei Engineering is primarily used to collect and identify user perceptual demands, as well as to guide how to match the correlation between user perceptual demands and robot design elements, which is consistent with our anticipated research content. Second, the existing research results primarily focus on the perceptual design of the overall robot form and do not address the perceptual design of industrial robot modules. Kansei Engineering (perceptual design) was implemented in the field of industrial robots, certain research outcomes were attained, and its theoretical and practical importance was demonstrated. However, there is still a great deal of work to be done with regard to Kansei Engineering for industrial robots, and a complete methodology has not yet been established. A pressing problem is how to integrate Kansei Engineering into the modular design of industrial robots. In this regard, against the backdrop of the trend of industrial robot modules gaining widespread use, how to conduct the configuration of industrial robot modules using Kansei Engineering in order to meet the increasing perceptual demands for industrial robot modules with a variety of forms and combinations has become a problem.

3. Materials and Methods

To effectively match industrial robot module configuration and a user’s perceptual demands, we propose a new method of industrial robot module configuration based on perceptual engineering, using a glass substrate transfer robot as an example. In this section, the research case, the research task, and the method’s implementation framework are sequentially presented.

3.1. Research Case

To demonstrate the practicability and reliability of the research methodology, we chose a glass substrate transfer robot for empirical analysis. There are two primary factors: this type of robot is mainly used for transferring ultra-thin and fragile glass substrates, which are key pieces of equipment for LCD, OLED, mini/micro-LED, and other display industries and have a broad market demand. Due to the display industry’s application requirements of a short production cycle and variable specifications, this glass substrate transfer robot can only be modularly designed to accommodate various production scenarios (according to statistics, this robot is expected to produce more than 100 types of specification types, which are designed to be applied to more than 20 production stations). Second, in order to contend with the fierce market competition and satisfy the user’s perceptual preference (BOE Technology Group Co., Ltd. Beijing, China), the style characteristics of the glass substrate transfer robot must be consistent with the brand image of the target user, i.e., the glass substrate module configuration solution has to reflect a “simple style”. Based on these two realistic requirements, we conducted research on the module configuration method of a glass substrate transfer robot based on Kansei Engineering and selected this robot as the application case for the research results.

3.2. Research Task

For the perceptual design problem of the glass substrate transfer robot module, we discovered, through a literature review and empirical investigation, that the ultimate objective of perceptual design is to scientifically match the relationship between user perceptual demands and product design features. To be specific, it examines the influential perceptual or emotional factors and how aesthetic preferences can be influenced in product design from a scientific cognitive and applied perspective, as well as how designers can effectively use or avoid such influences [19]. In light of this assertion, three tasks are assigned to the study of a Kansei Engineering-based method to configure glass substrate transfer robotic modules: identifying and transforming perceptual demands (Task 1), designing and coding robot modules (Task 2), and selecting module configuration solutions (Task 3). As the first task, Task 1 aims to clarify the relationship between user perception factors and the “simple style” of the robot, as well as to identify and extract some perceptional design points (these perceptual design points are utilized to assist designers in designing a simple glass substrate transfer robot). Based on these perceptual design points, the primary objective of Task 2 is to design and code the modules of the glass substrate transfer robot; the objective of Task 3 is to analyze and select the optimal glass substrate transfer robot from a number of module combinations.

3.3. Methodological Framework

Combining some of the most popular research methods and decision methods, we refined the above-described three research tasks, resulting in the methodological framework shown in Figure 1. The objective of the methodological framework is to systematically aid designers in applying Kansei Engineering to enhance the perceptual design of glass substrate transfer robot modules. This methodological framework comprises three phases and eight implementation steps, and the research content, methods, and anticipated results for each step in the specific operation are described below.

• Identifying and transforming perceptual demands

This task provides guidelines for the perceptual design of the glass substrate transfer robot module, which is accomplished in four steps.

Step 1: Collect and screen sample images. Providing tangible examples (physical objects, images, or videos) can facilitate the tracking of the fuzzy thinking in users’ minds when researching their perceptual demands. For this purpose, we conducted a common design survey, using keywords such as “transfer robots” and “simple style” to collect images and product descriptions of industrial robots from robotics brand websites, news articles, and the literature. The images were screened and clustered by our team (data screeners and industry experts). The specific process and results are described in Section 4.1 (1).

Step 2: Collect and analyze perceptual vocabulary. Perceptual vocabulary serves as a medium for expressing emotional imagery or aesthetic preferences, which is key to guiding industrial robot design from a perceptual perspective. For this purpose, we collected extensive information from users by using crawler tools, questionnaire research, and telephone interviews, considering the actual situation of the research data (sources and quantities). Then, as it is difficult to manually analyze a large number of user reviews, we used a combination of natural language processing (NLP) techniques and the expert judgment method [26]. The specific process and results are described in Section 4.1 (2).

Step 3: Measure the correlation between perceptual vocabulary and sample images. This study aimed to clarify the characteristics of users’ preferences regarding the appearance of industrial robots under their simplicity in terms of style. For this purpose, we measured the correlation between the perceptual vocabulary and the sample images using the semantic differencing method (SDM), in accordance with the experience of applying the literature [27]. As a measurement technique, semantic differencing is widely used to study users’ attitudes and perceptions. This is mainly achieved through the use of questionnaires (bipolar adjective scales), and its great advantage is that it is easy to administer and can be evaluated relatively quickly. The specific process and results are described in Section 4.1 (3).

Step 4: Summarize the key points of the robot’s perceptual design. The purpose of this study is to formulate a clear and effective perceptual design point that can be used to guide the design of industrial robot modules. To achieve this, we constructed a mapping relationship between user perceptions and glass substrate transfer robot design features using a hierarchical inference approach in conjunction with the user preference characteristics provided in Step 3 and then summarized the perceptual design points of the glass substrate transfer robot module. The specific process and results are described in Section 4.1 (4).

2.

Designing and coding robot modules

This task provides a physical reference for the perceptual design of the glass substrate transfer robot module, which is accomplished in two steps.

Step 5: Module division. As the initial step in the modular design process, the types of robot modules must be classified. To achieve this, we divided the glass substrate transfer robot into three types of intermediate modules based on their functions: auxiliary modules, joint modules, and functional modules. By combining expert recommendations and the structural features of the robot, we further subdivided and named the basic modules and sub-modules. The specific process and results are described in Section 4.2 (1).

Step 6: Module design and coding. Based on the perceptual design points provided by Step 4, we organized our engineering team to design the appearance of the glass substrate transfer robot module by combining the performance parameter conditions of motion, drive, communication, and control. Additionally, we set up a coding scheme to facilitate the configuration of the preferred glass substrate transfer robot module: Name and serial number of the base module_shape feature_color feature_material feature. The specific process and results are described in Section 4.2 (2).

3.

Selecting module configuration solutions

This task provides an application strategy for the perceptual design of the glass substrate transfer robot module, which is accomplished in two steps.

Step 7: Orthogonal design. According to the module combination rules, the 14 submodules provided in this study (see Table 4) result in 128 combination scenarios, which is significantly greater than the maximum number that can be manually evaluated (the maximum number of scenarios that can be effectively manually evaluated is generally considered to be 20, and there is also a view that the maximum number is 9). Evidently, it is difficult to select an optimal combination of these numerous module configuration options (that also meets the user’s perceptual demands). To improve the effectiveness of the evaluation of modular configuration solutions, it is logical to select a representative sample of modular combination solutions for evaluation from a large number of modules. To achieve this, we utilized an orthogonal design in SPSS (a data analysis software) to select a few representative scenarios from 128 configurations for evaluation, based on previous research [28,29]. The specific process and results are described in Section 4.3 (1).

Step 8: Conjoint analysis. Based on user decision-making processes, a conjoint analysis can identify which robot modules have the greatest influence on user preferences, as well as which configurations of modules are most popular. This can also determine whether the existing configurations are suitable and how they can be enhanced. Based on the results of the conjoint analysis, the designer can develop a configuration strategy for the robot module that maximizes the user’s perceptual demands. For this purpose, based on the literature [30], we evaluated the orthogonal design-screened module configuration options using conjoint analysis. The specific process and results are described in Section 4.3 (2).

4. Process and Results

In this section, we will detail the process and results obtained by the above-mentioned methods and steps. The section can be divided into three main parts, as follows.

4.1. Identifying and Transforming Perceptual Demand for Glass Substrate Transfer Robots

(1)

Collect and screen sample images

In accordance with Step 1, described in Section 3.3, this subsection describes the procedure for collecting and screening sample images, as well as the obtained results. First, we obtained a total of 400 robot images after three searches. Subsequently, we organized two groups of data-screeners (10 people, randomly assigned to 5 individuals in each group, with similar experience levels) in order to eliminate images with cluttered and blurred backgrounds through cross-screening and retained 103 images. Following this, we invited two groups of experts (six people, randomly assigned to three people in each group with the same level of experience) to remove images that did not reflect the study topic by cross-screening, and 47 valid images were retained. After that, the above two groups of experts were asked to cluster the 47 sample images using the KJ method (a method used to collect and integrate questions), and the 47 images were divided into 10 classes, each representing a typical robot case. Finally, we selected one case image from each category for further study; Figure 2 shows the final sample images. To eliminate distracting factors from users’ perceptual ratings, the background of the images was uniformly set to white, the images were of the same size (10 cm in length, 5 cm in width) and resolution (300 dpi), and the 10 sample images were numbered from Sample 1 to Sample 10.

(2)

Collect and analyze perceptual vocabulary

In accordance with Step 2, described in Section 3.3, we report on the process of collecting and analyzing the perceptual vocabulary, as well as the obtained results. First, 408 comments on the perceptual demands of robots were collected using crawler tools, questionnaires, and telephone interviews. Second, the information in the comments was pre-processed, word separation and high-frequency word extraction were obtained using natural-language processing, and two groups of experts were invited to de-duplicate, delete meaningless words, categorize, merge synonyms, and match antonyms on the comments after natural language processing. Finally, we were able to obtain six sets of perceptual vocabularies for industrial robots under the three appearance features of shape, color, and material.

Table 1. Six groups of perceptual vocabulary.

Appearance Features	Code Name	Vocabulary Groups	Definition
Shape	X1	Slender–Stout	Slender appearance—thick shape, heavy, large size
	Y1	Hard–Soft	Strong but not heavy—light and nimble, arms with curves
Color	X2	Simple–Rich	Simple colors, no more than three colors, avoiding color diversity and mixing colors
	Y2	Colorful–Grayish	Bright color and high saturation—low color saturation
Material	X3	Non-metallic–Metallic	Metal texture, hard material—non-metallic texture, visually soft material
	Y3	Rough–Smooth	No raised and depressed surfaces, complete body shell—bare wires and seams, rough in vision

shows information about the contents, coding modalities, and conceptual definitions of six groups of perceptual vocabulary. The results indicate that users frequently use "Slender–Stout" and "Hard–Soft" to describe robot shape features, "Colorful–Grayish" and "Simple–Rich" describe color features, and "Non-metallic–Metallic" and "Rough–Smooth" describe material features.

(3)

Measure the correlation between perceptual vocabulary and sample images

In accordance with Step 3, described in Section 3.3, this subsection describes the process used to measure the correlation between the perceptual vocabulary and the sample images, as well as the obtained results. The process was as follows: we developed a semantic differential questionnaire (an adjective scale with bipolarity) to measure the correlation scores between the 10 sample images (Figure 2) and the six sets of perceptual vocabularies (Table 1). The questionnaires were sent via email and distributed offline, and we ultimately collected 55 valid questionnaires (42 males and 13 females, with a mean age of 26.7, all indicating familiarity with the content of the questionnaire). As shown in

Table 2. Correlation scores between perceptual vocabulary and sample images.

Appearance Features	Perceptual Vocabulary	Sample Number										AVERAGE VALUE	Cronbach’s Alpha Coefficient
		1	2	3	4	5	6	7	8	9	10
Shape	X1_Slender–Stout	0.31	−0.2	0.02	−0.33	−0.42	−0.36	−0.24	−0.45	0.57	−0.27	−0.137	0.853
	Y1_Hard–Soft	−1.13	−0.6	−1.13	−0.38	−1.44	−0.33	1.73	−1.65	2.11	2.18	−0.264	0.876
Color	X2_Simple–Rich	−2.53	−0.51	0.15	−1.78	−1.33	−2.71	−1.73	−1.62	−1.05	−2.44	−1.555	0.879
	Y2_Colorful–Grayish	−2.35	−0.58	−1.04	−1.58	1.31	1.44	−1.71	−2.16	2.24	1.24	−0.319	0.886
Material	X3_ Non-metallic–Metallic	1.16	−1.6	1.18	1.02	1.11	1.4	1.6	1.62	2.11	1.05	1.065	0.843
	Y3_Rough–Smooth	1.47	1.73	1.2	1.35	0.33	1.51	1.91	1.56	2.57	−1.56	0.861	0.866

Likert scale score: −3 indicates extremely left correlation; −2 indicates very left correlation; −1 indicates general left correlation; 0 indicates no correlation; 1 indicates general right correlation; 2 indicates very right correlation; and 3 indicates extremely right correlation.

, the Cronbach’s alpha coefficients for all six groups of perceptual vocabulary were above 0.8, indicating the questionnaire’s high reliability. Furthermore, based on the data in Table 2, we plotted the distribution of each of the 10 sample images in accordance with the three appearance characteristics of shape, color, and material in order to more intuitively visualize the relationship between the perceptual vocabulary and the sample images (see Figure 3).

As presented in Figure 3, the 10 sample images are labeled in a four-dimensional quadrant. For example, Sample 1, under stylistic features, scores (−1.13, 0.31) on “X1_hard–soft” and “Y1_slender–stout”, falling into the second quadrant. According to the distribution (density and trend) of the 10 sample images in the four quadrants, we were able to determine the user preference characteristics for the three appearance features of shape, color, and material, as shown in

Table 3. User preference characteristics for the industrial robot’s appearance in a simple style.

Appearance Features	Characterization of the Scatter Plot (see Figure 3)	User Perceptual Preferences
Shape	10 samples are concentrated around the vertical midline, with 7 samples to the left of the vertical midline and 7 samples to the bottom of the horizontal midline	Users believe that slender, hard shape features best represent the simple style, but users prefer a hard shape
Color	9 samples are distributed to the left of the vertical midline, and all 10 samples are evenly distributed on either side of the horizontal midline	Under the simple style, users have no obvious color tendency for industrial robots, and color matching should be simple
Material	8 samples are distributed in the first quadrant (top right corner)	The user prefers a strong sense of metal, and the surface is smooth and has delicate material characteristics

. The results show that users preferred a slender and hard d shape, which best represented the simple style; users had no obvious color preference for industrial robots, and they preferred color combinations that are simple. A strong metallic texture is preferred by users, who consider materials with smooth and delicate surfaces to be the best representations of the simple style.

(4)

Summarize the key points of the robot’s perceptual design

In accordance with Step 4, described in Section 3.3, this subsection describes the process and results of summarizing the robot’s perception design points. In Figure 4, we present the mapping relationship between the user’s perceptual demands and the robotic features calculated by hierarchical inference. To be specific: the first column, the goal layer, represents the ultimate objective; the second column, the object layer, represents the object to be implemented to achieve the ultimate objective, which is the starting point; the third column, the perceptual layer, specifies the perceptual characteristics of the implemented object; the fourth column, the representation layer, identifies the specific perceptual demands; the fifth column, the mechanical object layer, matches the specific mechanical design object; and the sixth column, the design element layer, is used to interpret the implementation of the robot perceptual design strategy. In addition, based on the study presented in Figure 4, we summarize three perceptual design points for glass substrate transfer robots: to design a glass substrate transfer robot with a simple style, the design elements in the shape features are a streamlined drive shaft, a lengthwise expanded body structure, an integrated body structure, and a hidden plug; furthermore, simple color combinations and large theme colors can be used in the color features. The intended material features also include metal materials and polishing or painting surface treatment in order to achieve a smooth surface finish.

4.2. Designing and Coding Glass Substrate Transfer Robot Modules

(1)

Module division

In accordance with Step 5, described in Section 3.3, Figure 5 shows the results of the module division for the glass substrate transfer robot. As shown in Figure 5, the glass substrate transfer robot module is subdivided into three types of intermediate modules (the second column): auxiliary modules, joint modules, and functional modules. Intermediate modules are subdivided into five types of basic modules (the third column): base, turn shaft, lift shaft, arm, and end-effector (fork). The basic modules contain various types of sub-modules (the fourth column), and these sub-modules are indivisible modules in the glass substrate transfer robot module. These sub-modules are indivisible modules of the glass substrate handling robot, and different sub-modules can be combined to create various robot types that meet the flexible and diverse production requirements.

(2)

Module design and coding

In accordance with Step 6, described in Section 3.3,

Table 4. Design and coding of the glass substrate transfer robot module.

Basic Module	Design and Coding of Sub-Modules
Base	Base 1_Hard_Grayish_ Smooth and Metallic (abbreviation: B1)		Base 2_Slender_Grayish_ Smooth and Metallic (abbreviation: B2)
Turn shaft	Turn shaft 1_ Slender_ Grayish_ Smooth and Metallic (abbreviation: TS1)	Turn shaft 2_ Slender_ Grayish_ Smooth and Metallic (abbreviation: TS2)	Turn shaft 3_ Slender_ Grayish_ Smooth and Metallic (abbreviation: TS3)	Turn shaft 4_ Hard_ Grayish_ Smooth and Metallic (abbreviation: TS4)
Lift shaft	Lift shaft 1_ Slender_ Grayish_ Smooth and Metallic (abbreviation: LS1)	Lift shaft 2_ Slender_ Grayish_ Smooth and Metallic (abbreviation: LS2)	Lift shaft 3_ Slender_ Grayish_ Smooth and Metallic (abbreviation: LS3)	Lift shaft 4_ Hard and Slender_ Grayish_ Smooth and Metallic (abbreviation: LS4)
Arm	Arm 1_Slender_ Grayish_ Smooth and Metallic (Abbreviation: AR1)		Arm 2_Slender_ Grayish_ Smooth and Metallic (Abbreviation: AR2)
Fork	Fork 1_Slender_ Grayish_ Smooth and Nonmetallic (abbreviation: F1)		Fork 2_ Hard _ Grayish_ Smooth and Nonmetallic (abbreviation: F2)

presents the design results and code names for the glass substrate transfer robot module.

4.3. Selecting Configuration Options for Glass Substrate Transfer Robots

(1)

Orthogonal design

In accordance with Step 7, described in Section 3.3,

Table 5. Modular configuration options for glass substrate handling robots based on orthogonal design.

Serial Number	Configuration Scheme	Base	Turn Shaft	Lift Shaft	Arm	Fork
1	1	B2	TS3	LS3	AR2	F1
2	2	B1	TS4	LS1	AR2	F2
3	3	B1	TS3	LS4	AR1	F2
4	4	B1	TS3	LS2	AR2	F1
5	5	B2	TS4	LS2	AR1	F1
6	6	B2	TS1	LS2	AR2	F2
7	7	B2	TS3	LS1	AR1	F2
8	8	B1	TS1	LS3	AR2	F2
9	9	B2	TS2	LS1	AR2	F1
10	10	B2	TS4	LS4	AR2	F2
11	11	B1	TS2	LS2	AR1	F2
12	12	B1	TS1	LS1	AR1	F1
13	13	B1	TS4	LS3	AR1	F1
14	14	B2	TS1	LS4	AR1	F1
15	15	B2	TS2	LS3	AR1	F2
16	16	B1	TS2	LS4	AR2	F1

presents the screening results for the orthogonal design. As a result of the orthogonal design, 16 representative combinations were selected out of 128 possible configurations, thereby reducing the number of combinations that users must evaluate. Additionally, in order to make comparing and sorting the 16 combination scenarios easier for the user, we used 3D modelling software (C4D) to plot the effect of the 16 combinations in Table 5, as shown in Figure 6.

(2)

Conjoint analysis

In accordance with Step 8, described in Section 3.3, this subsection describes the process and results of a conjoined analysis based on the preferential selection of a modular system for transfer glass substrates. In accordance with the conjoint analysis experiment, 35 target users (28 males and 7 females) ranked the 16 module combination options (see Figure 6) according to their preferences (the further left the configuration option ID number is, the more preferred it is by the user). Then, we calculated the ranking results using SPSS conjoint analysis for the 35 target users. The results are shown in

Table 6. Results of conjoint analysis.

Basic Module	Importance (%)	Submodules	Utility Value	Standard Deviation	Internal Validity
Base	7.267	B1: Base 1_Hard_Grayish_ Smooth and Metallic	0.225	0.310	Kendall’s tau = 0.864 * p-value = 0.007 Notes: * = 5% significance level
		B2: Base 2_Slender_Grayish_ Smooth and Metallic	−0.225	0.310
Turn shaft	35.040	TS1: Turn shaft 1_ Slender_ Grayish_ Smooth and Metallic	1.081	0.538
		TS2: Turn shaft 2_ Slender_ Grayish_ Smooth and Metallic	1.394	0.538
		TS3: Turn shaft 3_ Slender_ Grayish_ Smooth and Metallic	0.294	0.538
		TS4: Turn shaft 4_ Hard_ Grayish_ Smooth and Metallic	−1.981	0.538
Lift shaft	31.129	LS1: Lift shaft 1_ Slender_ Grayish_ Smooth and Metallic	−1.744	0.538
		LS2: Lift shaft 2_ Slender_ Grayish_ Smooth and Metallic	1.156	0.538
		LS3: Lift shaft 3_ Slender_ Grayish_ Smooth and Metallic	0.506	0.538
		LS4: Lift shaft 4_ Hard and Slender_ Grayish_ Smooth and Metallic	0.394	0.538
Arm	12.437	AR1: Arm 1_Slender_ Grayish_ Smooth and Metallic	−0.394	0.310
		AR2: Arm 2_Slender_ Grayish_ Smooth and Metallic	0.394	0.310
Fork	14.127	F1: Fork 1_Slender_ Grayish_ Smooth and Nonmetallic	0.650	0.310
		F2: Fork 2_Slender_ Grayish_ Smooth and Nonmetallic	−0.650	0.310

Note: a cell with a darker color in the same column indicates that the data are more important.

: column 1 represents the robot’s base modules; column 2 indicates the relative influence of each base module on the user’s decision (how closely it conforms to the simplicity style); column 3 shows the submodules included in each base module; the utility value in column 4 indicates the degree to which each sub-module influences the user’s decision, with a higher value indicating that the sub-module is more important if it has a greater influence on the user’s perception of the robot’s simplicity; column 5 is the standard error; in column 6, the internal validity is represented as a correlation between the measured preferences of the target user and the estimated preferences of the statistical model, and it is used to determine the reliability of the analysis results, which can only be interpreted if the reliability meets the requirements of joint analysis.

According to Table 6, Kendall’s tau = 0.864 and p = 0.007, referring to the relevant usage experience [31] and thus suggesting that the results of this conjoint analysis have a reasonable degree of reliability. According to the results, the turn shaft module (35.040%) is the most important factor affecting the simplicity of the glass substrate transfer robot, followed by the lift shaft module (31.129%). The fork module (14.1127%) is in third place, followed by the arm module (12.437%) and, finally, the base module (7.267%). In addition, based on utility values (see column 4 in Table 6), we identified that B1, TS2, LS2, AR2, and F1 were the main submodules affecting the simplicity style of the robot, so we combined these submodules to obtain the best module configuration scheme, as in Figure 7.

In addition, the modular combination of glass substrate transfer robots shown in Figure 7 has already been mass-produced and received positive user feedback. It is important to note that combinations other than those in Figure 7 are also in production, and these perceptually designed modular combinations are better suited for the aesthetic preferences of users than those of competing robots, and they are gaining an advantage in the market.

5. Discussion

In this section, we discuss the main ideas, methodological strengths, research contributions, and limitations of this study.

5.1. Summary and Interpretation of Results

This study’s main purpose is to identify and confirm how the appearance factor in industrial robots contributes to users’ aesthetic preferences and how this can be matched to the features of industrial robot modules. In this regard, the study has revealed the following results.

(1)

According to the study, the shape, color, and material characteristics of industrial robots were the most important factors affecting users’ perceptions, which is consistent with the literature ([4,19]). Specifically, we discovered that users frequently used “hard–soft” and “slender–stout” to describe the robot’s shape; “colorful–grayish” and “rich–simple” to describe the robot’s color; and “smooth–rough” to describe the robot’s material.

(2)

As seen in Figure 3, seven samples (7/10) are concentrated to the left of the vertical center line in terms of shape features, indicating that users consider the “compact and slender” robot shape features to be the most representative of the simple style. This result confirms the literature findings [4], which suggest that “the robot’s design should be more compact, slender, integrated, and lightweight in terms of application and function.” Moreover, seven samples (7/10) were located below the horizontal midline, suggesting that “hard” was an important factor influencing user preference, and the density of distribution (all 10 samples were concentrated around the vertical midline) indicated that “hard” was more popular than “slender”.

(3)

As shown in Figure 3, nine out of ten samples (9/10) are distributed to the left of the vertical center line in terms of color features, which suggests that robots with simple color schemes are more appealing to users and create a “simple style” for robots. This result is supported by the literature [21], which suggests that industrial robots, like industrial machinery, should aim for simplicity in their color schemes. We can explain this phenomenon based on the literature findings [32]: the pastel colors represent a relatively plain robot body, which conveys a sense of stillness. In addition, we suggest that industrial robot theme colors be chosen from a low-stimulus affinity color, despite the fact that Figure 3 does not reveal a significant color preference (10 samples, evenly distributed). This recommendation is based on the literature [4], which states that “soft and moderate colors for industrial robots can reduce the stress of workers and reduce the occurrence of accidents”.

(4)

As shown in Figure 3, eight samples (8/10) are concentrated in the first quadrant (top right corner) in terms of material features, indicating that users prefer a strong metallic feel and believe that a smooth and delicate surface material feature best reflects the “simple style”, as evidenced by the design advice provided by industry experts in the preliminary research.

(5)

Figure 4 reports the mapping relationship between user perceptual demands and robot design features, from which we can summarize the key points of perceptual design for industrial robot modules. The results indicate that, when designing robot modules with simple style features, streamlined drive shafts, lengthwise expanding body structures, integrated body structures, and hidden plugs can be considered shape features; large theme colors and simple color schemes can be used as color features; and smooth surface textures can serve as material features as much as possible.

(6)

Table 6 reports the relative importance of the five base modules in influencing the aesthetic features, and the results show that the steering axis module (TS) and the lifting axis module (LS) are key influences on the stylistic features of the glass substrate handling robot. The results of this study confirm the literature ([21,22]), which states that “the waist and base of an industrial robot are the first areas of observation for the user” and that these two types of modules should be specifically designed.

5.2. Methodological Advantages and Contributions

It has been suggested in the literature ([4,19,21,22,33]) that, when differentiation becomes difficult on the basis of “functional utility” and “technology” alone, perceptual or emotional design will inevitably be one of the most important factors differentiating robots from the competition. Nevertheless, through a review of the existing literature on perceptual engineering, we found that perceptual design methods for industrial robots have not yet been explored in depth, and a comprehensive methodological framework has not yet been developed, particularly since how to incorporate perceptual design into the modularization of industrial robots remains unclear. In order to fill this gap, we propose a Kansei Engineering-based method for the configuration of glass substrate transfer robot modules, which has the following advantages and contributions.

(1)

This method provides a complete framework for the perceptual design of industrial robots as compared to existing research methods. Specifically, the current study develops a multistage perceptual design framework for industrial robots, whereas previous research has only focused on a single stage of perceptual design [19,21,22] (constructing a perceptual imagery space for industrial robots). Using Figure 1, we analyzed perceptual demands, extracted design elements, generated design concepts, evaluated combination solutions, and implemented a complete perceptual design process (from design to evaluation).

(2)

This method broadens the scope of the theoretical study of Kansei Engineering in the field of the modular design of industrial robots. This study explains the mechanisms by which the appearance features and modular components of industrial robots influence the aesthetic preferences of users in the context of robotics and robot modular design. This study facilitates the study and application of Kansei Engineering in industrial robot modular design, thereby broadening the theoretical scope of Kansei Engineering in previous industrial robot research.

(3)

This method has been shown to improve the uniformity of style in the configuration of industrial robot modules. Modular configurations of industrial robots have the disadvantage of “a thousand robots, a thousand faces” at present, which does not satisfy the user’s perceptual demands for uniformity, consistency, and standardization. However, previous research has focused on the overall design of industrial robots, and this method has resulted in the problem of “one module configuration equals one redesign, and a new style is created”, which not only reduces the timeliness of module design but also easily leads to stylistic differences between module configurations. In contrast to previous research, we decompose the influence of the appearance characteristics of industrial robots and module components on users’ aesthetic preferences and propose a configuration path for industrial robot modules based on unified style characteristics, enabling an “a thousand robots, one face” style design for various module configurations.

5.3. Limitations and Further Research Needs

While the present study has provided some positive explorations of perceptual design methods for industrial robot module configurations and applications, as well as explanations of their contributions and strengths, it is also important to highlight the limitations of these studies, including their subjects, methods, and results, as well as their future research needs.

(1)

It is important to note that our research is limited to the perceptual design and practical application of a glass substrate transfer robot module. The findings of this study could serve as a guide for the design of a transfer robot of the same type, but they should be applied with caution to other types of robots. For instance, the perceptual design of service robots, social robots, and collaborative robots should be supplemented by research on the perceptual demands of safety, trust, and natural interaction, as opposed to the focus of this study, which is on the appearance of industrial robots. Essentially, the closer a robot is to a human, the more perceptual and emotional factors need to be considered. In the future, further research and improvements will be required, based on experience.

(2)

During the collection and analysis of the perceptual vocabulary (see Section 4.1(2)), we did not verify whether all reviewers were familiar with the working environment of transfer robots. In this regard, we were only able to determine and select the relevance of the review content to transfer robots, and we organized two groups of industry experts to supplement the findings in this section. Despite the positive effect that this method of bulk information collection had on the design practice of this study and on saving analysis time, the reliability of the sources of these perceptual terms should be considered in future research—particularly in theoretical methods—in order to prevent inaccuracies in theoretical studies due to generalized information.

(3)

A note of caution should be made: the application of the above findings is limited to a “Simple Style” due to the design requirements of the target company, and additional research should be conducted on research hypotheses that extend beyond this type of perceptual style.

6. Conclusions

To address the research gap in Kansei Engineering regarding the modular design of industrial robots, this study proposed and validated a Kansei Engineering-based modular configuration method for industrial robots, using a glass substrate transfer robot as a model. The methodology is based on the main task of “matching the relationship between the user’s perceptual demands and the product design features” and focuses on identifying and transforming perceptual requirements, designing and coding industrial robot modules, and selecting industrial robot module configurations. The results of this study explain the mechanism by which the aesthetic preferences of users are influenced by the appearance of industrial robots and their modular components, thereby resolving the “a thousand robots, a thousand faces” problem of previous industrial robot modular configurations from the perspective of uniform stylistic features to meet business and application requirements. This study constructed a comprehensive perceptual design framework for industrial robots, effectively reducing the “distance” between the designer and the design object. Additionally, its theoretical and applied research in the field of modular design for industrial robots provides a new reference for the perceptual design of other types of robot modules.