Hierarchical modeling of mainstream functional requirements for intelligent cockpits

Abstract

Background

Potential functional user needs have become more valuable and difficult to satisfy as the problem of homogenization of intelligent cockpit (IC) functions has intensified.

Objective

To address the mismatch between IC market offerings and user demands for personalized functions, this study constructed an empirically validated functional hierarchy model for IC design optimization.

Methods

Thirty-one functional requirements were identified by combining with data on shortcomings in market-installed IC functions through comprehensive analyses of 31 IC products and data collected via web crawlers along with semi-structured interviews with 12 automotive ergonomics experts (mean experience = 11.5 years). These requirements were reduced to five key categories using principal component analysis with orthogonal rotation by varimax rotation (Kaiser–Meyer–Olkin measure = 0.922, Bartlett's sphericity p < 0.001). Functional weights were evaluated via the analytic hierarchy process (AHP) with a consistency ratio (CR < 0.1).

Results

The intelligent interconnection mode emerged as the highest-priority category (global weight = 0.4225), with head unit functions significantly outperforming other functions (M = 0.1573, p < 0.01). Despite high consumer demand, the sentinel mode showed a low installation rate (20%) in the Chinese market. The phone-vehicle interconnectivity mode was considered the least essential (priority number 5).

Conclusion

The model quantified functional priorities using human-centered metrics, reconciling technical feasibility with user expectations.

Keywords

ergonomics ergonomics in manufacturing ergonomics tool design and evaluation human factors and sustainable development product design psychophysiology in ergonomics

Introduction

The rapid advancement of intelligent and internet-connected vehicle technologies has established intelligent cockpits (ICs) as critical components of modern automotive design and intelligent driving systems. An IC is an integrated system that combines advanced hardware and software with intelligent and internet-connected capabilities, enabling real-time perception, decision-making, and adaptive learning within the vehicle cabin. As user demands continue to rise, IC functionality is steadily expanding, increasing their popularity. However, short development cycles, rapid knowledge updates, and the dynamic evolution of user requirements in the IC domain pose significant challenges to maintaining high levels of user satisfaction. Consequently, the accurate and timely acquisition of user needs remains a crucial task.¹

User needs form the foundation of IC functional layout design. The extent to which these needs are met and the effectiveness of the layout significantly influence the market competitiveness of IC systems.¹ Various methodologies have been employed to capture these needs, including: in-depth interviews,² questionnaire surveys,³ and data mining techniques.⁴ Additionally, the installation rate of specific IC functions serves as a key indicator of user acceptance, directly affecting research and development (R&D) priorities and supply chain planning. Consequently, it has emerged as a critical strategic variable in the product development process.⁵ A balanced consideration of user demand data and function installation rates is therefore essential for making scientifically sound and economically viable decisions in the design and deployment of intelligent cockpit systems.

However, most existing research has focused on optimizing and exploring functional details within IC systems,^6–12 while neglecting overall functional layout and planning. Furthermore, many existing studies on IC design have concentrated primarily on evaluating individual functions or performance indicators, producing context-specific conclusions that vary with the research perspective.

Recently, the IC function hierarchy model has emerged as an effective framework for quantifying and prioritizing user needs, attracting growing interest in both academic and industrial research.^13–16 This model enables a systematic classification of in-vehicle functions based on their relative importance, facilitating more informed design and development decisions. Several scholars have applied this model to assess the impact of functional features on consumer behavior and product design. For example, Zhou and Li¹⁷ categorized smart car features, noting that electric and smart attributes strongly influence the adoption of new energy vehicles, while driving range had a lesser impact. Dong¹³ examined the effect of mainstream IC functions on purchase intentions, identifying voice assistants and smart rearview mirrors as highly influential. Chang¹⁴ highlighted head units, voice and gesture interactions, head-up displays (HUDs), electronic mirrors, and smart seats as design priorities based on user demand data. Mihale-Wilson et al.¹⁵ found that users prioritize driving-related information over entertainment, prefer voice assistants, and consider system errors more critical than input convenience. Over time, fuel efficiency and seat comfort have increased in importance, whereas interior layouts and prices have declined. Lai, Lin¹⁸ identified voice interaction, remote services, in-car gaming, and auto-parking as key features affecting user satisfaction in IC systems. In a methodologically comprehensive study, Zhu, Sun¹⁹ involved 50 experienced drivers (with more than three years of driving experience) to construct an IC seat evaluation model by integrating the hierarchical analysis method, Delphi method, and d-number theory. Their analysis revealed that seat safety experience has the greatest impact on the overall user experience of the intelligent cockpit seat, followed closely by seat comfort.

Overall, previous research has three main limitations. First, an overreliance on subjective data undermines methodological rigor. Second, studies often focus on a static set of core functions, neglecting the integration of emerging features and thereby failing to capture their dynamic interaction with user demands. Third, the predominant emphasis on user needs overlooks the critical market metric of functional adoption rates, introducing potential bias. Additionally, user behavior has been shown to significantly influence these outcomes, resulting in varied interpretations and findings.^3,8 Despite these efforts, it remains unclear which functions or functional categories best align with user requirements, and the relative importance and hierarchical prioritization of IC functions have not yet been explored in an objective and systematic manner. Bridging this gap requires a scientifically grounded model that integrates user experience, demand, and actual installation data to hierarchically prioritize IC functionalities.

This study aimed to identify the functional categories that best align with user needs in ICs. To achieve this, a mainstream IC function library was first constructed based on expert knowledge and industry insights. The identified functions were then classified using a questionnaire. Principal component analysis (PCA) was employed to reduce dimensionality, identify underlying patterns, and group functions into distinct categories. Subsequently, the analytic hierarchy process (AHP) was applied to evaluate the priority and importance of each function by integrating user demand data with installation rate statistics. By establishing a systematic, data-driven framework for hierarchically prioritizing IC functions, this research aims to bridge the critical gap between user needs, market realities, and design decisions, enhancing the IC user experience, increasing engagement with in-vehicle systems, improving R&D efficiency for automakers, optimizing feature usability and relevance, and reducing the risk of mismatches between design and user expectations.

Methodology

The research procedures in this study were conducted in accordance with the principles outlined in the Declaration of Helsinki²⁰ to ensure ethical integrity (No. TJUE-2024-139) and protect participant welfare throughout all stages of the research process.

This study employed a mixed-methods research design, integrating qualitative interviews and quantitative questionnaires to comprehensively capture both the subjective preferences and objective data related to IC functions. Methodological reporting strictly followed the guidelines for reporting mixed methods studies (GRAMMS²¹). Specifically, the semi-structured interviews conducted to construct the mainstream function library adhered to GRAMMS’ qualitative dimension requirements; the online questionnaire survey used to classify mainstream functions followed the quantitative dimension guidelines; and the data analysis for modeling the mainstream IC function hierarchy complied with the integration requirements specified in GRAMMS.

The methodological framework for developing the IC functional hierarchy model comprised three iterative stages, as shown in Fig. 1 : the user, functional, and physical layer.

Figure 1.

User-functional-entity hierarchy model.

In the user layer, a mainstream function library was established by identifying functions that align with end-user needs and market demands. This was achieved through a multi-source approach, including a literature review, web crawling of public automotive market databases, semi-structured interviews with industry professionals, and market research surveys.

In the functional layer, building on the function library developed in the user layer, a questionnaire survey was administered to frequent vehicle users, focusing on key evaluation dimensions, including functional practicality, human–machine interaction (HMI) compatibility, and willingness to pay for specific features. This process resulted in a core mainstream function library for the IC, reflecting the most relevant and preferred functions from a user-centric perspective.

In the physical layer, PCA was applied to screen and categorize the functions into distinct groups based on statistical patterns and relevance. Subsequently, the AHP was used to objectively evaluate and prioritize the importance of each function. Finally, the relative importance of each mainstream function was determined, and a comprehensive functional hierarchy model was constructed to systematically represent user priorities and functional relationships.

Building the mainstream function library of ic

The functional classification in this study was developed within a dual-aspect framework that aligns with the ISO 20282-2:2019²² accessibility standards. This framework was grounded in systematic market research, which included web crawling and semi-structured interviews with 12 domain experts from the Chinese Automotive Technology and Research Center Co. (CATARC).

Crawled data were collected from a range of Chinese public sources between September 2 and 20, 2024, including automotive review platforms (e.g., AutoNews²³ and Gasgoo Research²⁴), industry reports (e.g., from the China Passenger Car Association²⁵), and forums (e.g., PCauto Club²⁶). Data collection was conducted using Python and Selenium,²⁷ employing Chinese keywords translated into English as “intelligent cockpit,” “vehicle-computing integration,” “multimodal interaction,” “function adoption rate,” “user satisfaction” and “repurchase intention.” In addition to the crawled data, product updates from leading automakers’ official websites were also referenced to ensure the relevance and timeliness of the identified functions. By combining the crawled data with functional installation rates collected from September 23 to 30, 2024, a total of 31 mainstream functions were extracted and organized into a mainstream functional library.

Additionally, through purposive sampling,²⁸ semi-structured interviews were conducted with 12 CATARC experts on September 30, 2024. The sample included nine men and three women, with a mean age of 35 years and an average professional experience of 11.5 years. All participants had at least five years of experience in vehicle or IC design and research. Prior to the interviews, informed consent was obtained from all participants to ensure ethical compliance and data privacy. The sample size was guided by previous studies that included 10²⁹ and 11 experts.³⁰ In collaboration with CATARC designers, the interview questionnaire (Appendix 1) comprised six key questions addressing the following aspects related to mainstream IC features: current mainstream market features, estimated penetration rate, estimated level of user demand, technological maturity, user popularity, future development expectations, and additional insights. The interviews were conducted via telephone and lasted approximately 30–45 min. All collected data were used exclusively for internal research, and participants’ identities were anonymized throughout the entire research process. Participants were physically and cognitively healthy, and fully capable of understanding the interview objectives. No participants withdrew from the study.

This dual-method approach produced two distinct functional taxonomies: (1) user-perceived functional priorities and (2) market-derived feature prevalence. The categorization criteria were validated through expert consensus (see Appendix 2 for details).

Building on expert opinions, market penetration rates, and future development trends, 111 vehicle functions were analyzed in terms of their frequency of occurrence and expert evaluation scores, and subsequently integrated and refined to form the specific functions listed in Table 1, For definitions of the 31 mainstream functions listed, please refer to Supplementary table 1. The most recent market installation rates of these 31 functions were examined (see Table 1), providing empirical support for the scientific validity and robustness of the subsequent analyses and model construction.

Table 1.

Installation rates of 31 mainstream IC functions in the Chinese market.

Mainstream functions	Functional installed base	Mainstream functions	Functional installed base
Voice interaction⁴¹	78.80%	Head unit⁴¹	92.60%
Co-pilot screen⁴²	4.30%	Rear screen⁴³	2.10%
Gesture recognition⁴³	4.40%	In-vehicle facial recognition⁴⁴	15.00%
Multimodal interaction⁴⁵	10.00%	Digital key⁴⁶	33.35%
Remote control vehicle⁴⁷	74.00%	Mobile phone and car interconnection⁴¹	56.40%
Cross-screen connectivity between cell phone and vehicle systems⁴⁵	56.40%	Bluetooth connection⁴⁷	95.00%
WiFi connection⁴⁴	80.00%	360 Panoramic surround view⁴⁸	47.07%
Head-up display (HUD)⁴¹	10.40%	Sentinel mode⁴⁵	20.00%
Camera-monitor system (CMS)⁴⁹	1.60%	Car navigation function⁵⁰	80.10%
Wireless charging of cell phone⁵¹	44.10%	Intelligent trunk⁵²	42.10%
Car software application market⁵³	30.00%	Seat ventilation⁵⁴	2.70%
Seat heating⁵⁵	7.80%	Seat massage⁵⁶	9.03%
Seat position memory⁵⁷	30.71%	Large angle seat⁴⁵	20.00%
Ambient lighting⁵⁸	44.10%	Active noise reduction function³⁹	8.00%
Intelligent audio⁴⁵	20.00%	Independent sound zone⁴⁵	10.00%
Intelligent air condition⁴⁵	20.00%

Classification of mainstream functions

The 31 IC functions listed in Table 1 were rigorously validated through a questionnaire-based investigation (The sample questionnaire is shown in Appendix 3) to ensure their representativeness and relevance to user needs.

Participants

The investigation involved 722 certified drivers (mean_age = 34.3 ± 4.7 years, driving experience = 5.3 ± 3.1 years) with the following demographic and experiential characteristics. The participants pool, comprising individuals aged 18 to 60 years, was diverse in terms of educational background, place of residence, marital status, parenthood, car-purchasing budget, and current car ownership, ensuring a representative cross-section of the target population. Prior to the survey, each participant provided informed consent online in accordance with ethical research standards. The sample size was determined to achieve a 95% confidence level with a ± 3.78% margin of error (e), based on the total population of licensed drivers in China (N = 487 million in 2022). The margin of error was calculated using Cochran's formula for finite populations,³¹ as follows, ensuring the statistical validity and generalizability of the findings:

e = \sqrt{\frac{Z^{2} \cdot p \cdot (1 - p)}{n_{0}}}

(1)

where,

n_{0} \approx n = 676

, Z = 1.96 (corresponding to confidence level of 0.95), and p = 0.5 (a conservative default to maximize variance, ensuring the sample size is sufficient for all potential responses).

Questionnaire

The questionnaire used in this study employed a five-point Likert scale,³² and consisted of 39 validated items designed to evaluate functional utility and human-system compatibility (shown in Appendix 3). The survey was distributed via Wenjuanxing, a leading online survey platform in China, to ensure broad and convenient access for the target audience. Of the 39 items, nine were demographic questions aimed at collecting basic participant information, including residence, age, and gender. The remaining 31 items assessed participants’ opinions on the 31 core IC functions identified in this study. Each function was rated on the following scale: 5 = essential (must-have, should be standard); 4 = conditionally desirable (appropriate cost-benefit ratio, reflects willingness to pay); 3 = marginal utility (not justifying additional expenditure); 2 = unaware of the feature, and 1 = active rejection. The scale anchors were pretested with 15 participants to ensure the construct validity, and the Cronbach's α coefficient was 0.811, indicating acceptable internal consistency. The first page of the questionnaire provided participants with basic study information and a clear explanation of the purpose and scope of data usage, in line with ethical research practices. The survey remained open for a seven days, from October 25 to 31, 2024, allowing sufficient time for data collection.

Data acquisition

The data collection process yielded 722 initial responses, of which 46 were excluded following rigorous quality control protocols to ensure data validity and reliability. The exclusion criteria were as follows: 1) Random response detection: Two attention-check items were embedded in the questionnaire (e.g., “Select ‘Strongly Disagree’ for this item”) to identify and remove inattentive or careless responses; 2) Time-based screening: Responses completed in under 120 s were excluded, as such rapid completion suggested a lack of genuine engagement. After applying these criteria, the final analytical sample comprised 676 valid responses, representing an effective response rate of 93.6%. The demographic distribution of the final sample was consistent with the broader automotive consumer population in China.³³ The sample included 478 males (mean age = 34.50 years, SD = 8.88) and 198 females (mean age = 34.10 years, SD = 9.60). These demographics reflect a balanced and representative sample across genders and age groups, facilitating generalizable conclusions.

Before conducting hierarchical clustering based on cognitive workload metrics and task criticality indices, the response data underwent preliminary psychometric and statistical checks, including normality testing and reliability analysis: The Shapiro-Wilk test confirmed that the data distribution was approximately normal (p > 0.05), and the 31-item IC feature scale demonstrated high internal consistency, with a Cronbach's α of 0.916 (95% confidence interval: 0.901–0.930), well above the recommended threshold of 0.70 for group-level comparisons. For factor analysis assumptions, the Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy was 0.922, indicating strong suitability for factor analysis. Bartlett's test of sphericity was statistically significant (χ² = 4967.508, df = 231, p < 0.001), confirming sufficient correlations among items to justify factor extraction. These psychometric properties validate the robustness and reliability of the measurement instrument, supporting subsequent data analysis and model development.

Data preprocessing

PCA was employed in this study to reduce data dimensionality and identify the most informative underlying patterns among the 31 evaluation items.

Standardization of raw data. The datasets consisted of five indicator variables ( $x_{1}, x_{2}, \dots, x_{5}$ ) and 31 evaluation objects $(x_{1}, x_{2}, \dots, x_{31})$ . Let x_ij represent the value of the j^th indicator for the i^th evaluation object. The standardization process transforms the original data into a normalized matrix Y , defined as follows.

Y_{ij} = \frac{x_{ij} - {\bar{x}}_{j}}{s_{j}}, i = 1, 2, \dots, 31; j = 1, 2, \dots, 5

(2)

{\bar{x}}_{j} = \frac{\sum_{i = 1}^{31} x_{ij}}{31}, s_{j}^{2} = \frac{\sum_{i = 1}^{31} {(x_{ij} - {\bar{x}}_{j})}^{2}}{31 - 1}

(3)

Calculation of the correlation coefficient matrix $R = (r_{ij})_{p \times p}$ .

r_{ij} = \frac{\sum Y_{kj} \cdot Y_{kj}}{31 - 1}, i = 1, 2, \cdot \cdot \cdot, 31, j = 1, 2, \dots, 5

(4)

Calculation of the factor load matrix A. Using PCA, the corresponding eigenvectors ( $μ_{1}, μ_{2}, \dots, μ_{31})$ were first computed to represent the directions of maximum variance in the data. Typically, the cumulative variance explained by the principal components exceeds 70%, which is considered sufficient for dimensionality reduction and pattern extraction in most applications. In this study, a more stringent threshold of 80% was selected to ensure a higher data retention and greater representativeness of the reduced feature space. This decision was made to enhance the reliability and robustness of the subsequent analysis, particularly to capture the essential variability among the 31 evaluation objects.

u_{ij} = \frac{\sum_{j = 1}^{m} λ_{j}}{\sum_{j = 1}^{5} λ_{j}} \geq 0 .80,

(5)

then the factor loading matrix is

A = [\sqrt{λ_{1}} u_{1}, \sqrt{λ_{2}} u_{2}, \dots \sqrt{λ_{m}} u_{m}] .

(6)

Data analysis and results

Classification results of mainstream ic functions

Principal component analysis of mainstream IC functions

A PCA with varimax rotation was conducted on the 31 IC features using IBM SPSS Statistics 21 (v21.0.0, Armonk, NY, USA) to identify underlying latent dimensions and extract a parsimonious set of representative functions. The scree plot (Fig. 2) indicated that five principal components satisfied the Kaiser criterion (eigenvalues > 1), collectively explaining 56.91% of the total variance (component 1: 15.04%; component 2: 12.70%; component 3: 12.15%; component 4: 10.65%; component 5: 6.36%). Component extraction was further validated using parallel analysis, in which the eigenvalue of the fifth component (1.139) exceeded the 95^th percentile of Monte Carlo–simulated eigenvalues (1.09), confirming its statistical significance and retention in the final solution.

After varimax rotation, nine features were eliminated based on standard criteria: (a) cross-loadings > 0.40 on multiple components,³⁴ indicating ambiguous dimensionality, and (b) communalities < 0.50, suggesting insufficient variance explained by the extracted components. The final factor solution retained 22 core functions, with a mean communality of 0.569 (SD = 0.067) and rotated factor loadings ranging from 0.49 to 0.70, demonstrating adequate convergent validity and construct representation.

The PCA of the 22 IC features yielded five components that satisfied the Kaiser criterion (eigenvalues > 1). These components explained 15.04%, 12.70%, 12.15%, 10.65%, and 6.36% of the variance, respectively, with a cumulative variance contribution of 56.91%. This final factor solution accounted for over 50% of the total variance, exceeding the commonly recommended threshold for exploratory factor analysis (EFA) in human-machine systems research.³⁴ These results indicate that the extracted components adequately capture the underlying structure of the IC feature set, providing a statistically sound and interpretable foundation for further analyses.

Principal component factor rotation of mainstream IC functions

To achieve a simple structure and enhance the interpretability of the extracted factors, an EFA with varimax orthogonal rotation was performed. This rotation maximized the variance of factor loadings across variables, producing a clear and distinct loading pattern. Results indicated that 89% of the measured variables exhibited salient loadings (> 0.6) on their primary factors, consistent with Barbara G. Tabachnick³⁵ guidelines for identifying meaningful latent constructs. The factor extraction process converged in six iterations using a convergence criterion of 1 × 10⁻⁶, and all communalities exceeded the minimum threshold of 0.40, ensuring adequate item representation.³⁴ The final rotated solution retained five orthogonal factors (F1–F5), collectively explaining 56.91% of the total variance. Each factor demonstrated discriminant validity, as the differences in loadings between primary and secondary factors exceeded 0.3, supporting the distinctiveness of the underlying dimensions.

Naming and categorizing of mainstream IC functions

Factor analysis was applied to the 31 mainstream IC functions of the IC to systematically categorize them based on their psychometric properties and underlying constructs. This analysis identified five distinct factor categories. The functions within each category, along with their interpretations and groupings, are summarized in Table 2.

Figure 2.

Scree plot of 31 mainstream functions of the IC.

Table 2.

Mainstream functional categories derived from PCA of IC.

Factor	Function category name	Mainstream functions
Factor 1	Immersion features	Seat massage, large angle seats, ambient lighting in the IC, smart audio, separate sound zones, active noise cancellation
Factor 2	Intelligent interconnection functions	Voice interaction, head unit, digital key, remote control vehicle, car navigation, car software application market
Factor 3	Interactive experience functions	Co-pilot Screen, rear screen, gesture recognition, multimodal interaction
Factor 4	Cell phone-vehicle interconnection function	Mobile phone and car interconnection, cross-screen connectivity between cell phone and vehicle systems, bluetooth connection, WiFi connection¹
Factor 5	Security monitoring functions	Sentinel mode, camera-monitor system (CMS)

Modeling of mainstream IC function hierarchy

To systematically present the outcomes of the mainstream functional hierarchy division of the IC, the relative importance of the identified functional categories was evaluated and ranked based on their variance contributions and factor loadings. This evaluation provided a quantitative basis for understanding the priority and relevance of each category in the context of user demand and functional integration. Subsequently, a mainstream IC function hierarchy division model was developed to organize and structure these functional categories within a hierarchical framework. The model was specifically designed to establish a bidirectional mapping between the user demand perspective and the IC's functional hierarchy, thereby enhancing alignment between user expectations and system design. This approach improves the interpretability of the functional classification and supports decision-making in product development and feature prioritization.

Weight calculation based on the AHP

A total of 676 valid questionnaires were analyzed using SPSS Statistics 21 to determine the relative importance of the indicators. The AHP was applied to calculate the weights of the criteria in a structured and systematic manner.

Step 1: Judgment matrix construction

The Saaty's 1–9 point scale method³⁶ was used to construct the pairwise comparison matrix. This approach employs a numerical scale ranging from 1 to 9, including their reciprocals, to quantify the relative importance of the evaluated factors. The specific scale values and their corresponding definitions are summarized in Table 3, oviding a clear and consistent framework for conducting pairwise comparisons and deriving the indicator weights.

Table 3.

Hierarchical analysis nine level scaling scale.

Digital scale	Evaluation criteria
1	A is equally important compared to B
3	A is slightly more important than B
5	A is more important than B
7	A is clearly important compared to B
9	A is particularly important compared to B
2, 4, 6, 8	Midpoint
1/3, 1/5, 1/7, 1/9	B is more important than A

The judgment matrix for the criterion layer A is

A = [\begin{matrix} a_{11} & a_{12} & \dots & a_{1 n} \\ a_{21} & a_{22} & \dots & a_{2 n} \\ \dots & \dots & \dots & \dots \\ a_{n 1} & a_{n 2} & \dots & a_{nn} \end{matrix}]

(7)

where denotes the importance score of a given function relative to the other functions in the pairwise comparison, and n represents the serial number of the functions in a randomly assigned sequence.

Step 2: Hierarchical single ordering

In the hierarchical single-ordering phase, pairwise comparisons were conducted for all elements within the five function categories relative to a higher-level element from the preceding hierarchy level, based on their relative importance. This process produced corresponding weight values for each element—denoted as (w₁, w₂, …, w₅) — providing a quantitative measure of their significance within the hierarchical structure.

Step 3: Matrix consistency test

A consistency test was conducted to validate the logical coherence of the constructed pairwise comparison matrices. This test involved calculating the consistency ratio (CR) and comparing it with the random consistency index (RI) to determine whether the matrix met the acceptable consistency criteria in the AHP. A matrix is considered logically sound and suitable for further analysis if the CR value is ≤ 0.10, which is the commonly accepted threshold for consistency. The consistency test was performed using the following procedure:

The consistency index (CI) was computed using the following equation:

CI = \frac{λ_{\max} - n}{n - 1}

(8)

where,

λ_{\max}

is the maximum eigenvalue of the pairwise comparison matrix, and n is the number of elements being compared. The maximum eigenvalue

λ_{\max}

was derived from the following equation:

λ_{\max} = \frac{1}{n} \sum_{i = 1}^{n} \frac{{(Aw)}_{i}}{w_{i}}

(9)

The RI values used in the test were obtained from Saaty's simulations based on 1000 random pairwise comparisons, as summarized in Table 4.

Table 4.

Stochastic RI value results.

Matrix order n	1	2	3	4	5	6	7	8	9	10	11
$RI$	0	0	0.58	0.90	1.12	1.24	1.32	1.41	1.45	1.49	1.52

Note: RI means random index.

Table 4 Stochastic RI value results

The CR was then calculated as follows:

CR = \frac{CI}{RI}

(10)

A CR $\leq$ 0.10 indicates acceptable consistency, ensuring the judgments embedded in the matrix are reliable and rational.

Ranking results of mainstream IC function categories

After calculating the mean weight across all samples to determine the base weights of mainstream functions, the weight of each function was multiplied by its corresponding market installation rate, yielding the composite weights of individual mainstream functions, as summarized in Table 5.

Table 5.

IC mainstream functions installation rate.

Mainstream functions	Mean installation rate
Immersion features	20.683%
Intelligent interconnection functions	64.675%
Interactive experience functions	5.275%
Cell phone-vehicle interconnection function	60.350%
Security monitoring functions	5.000%

The final mean functional indicator weights were calculated as follows: intelligent interconnection functions, 42.250%; interactive experience functions, 18.272%; immersion features, 15.653%; security monitoring functions, 14.954%; and cell phone–vehicle interconnection function, 8.870%. These results met the AHP consistency criteria, with the CI = 0.094 < 0.1 and the CR = 0.084 < 0.1, indicating that the judgment matrix was logically consistent and the derived weights were reliable. Based on these mean functional indicator weights, the hierarchical ordering of the mainstream IC functions, from highest to lowest priority, was: intelligent interconnection functions, interactive experience functions, immersion features, security monitoring functions, and cell phone–vehicle interconnection function.

The priority ranking of each sub-function within its respective mainstream function category was also determined. This process enabled the calculation of the functional proportions of the 22 core mainstream functions, as detailed in Table 6.

Table 6.

Sub-function hierarchy ranking results of IC mainstream function categories.

Mainstream functional class	Mainstream functions	Functional percentage	Functional hierarchy sorting order
Intelligent interconnection functions	Voice interaction	5.413%	7
	Head unit	15.730%	1
	Digital key	2.776%	12
	Remote control vehicle	6.065%	6
	Car navigation	10.409%	3
	Car software application market	1.857%	16
Interactive experience functions	Co-pilot screen	2.492%	14
	Rear screen	1.180%	19
	Gesture recognition	3.575%	11
	Multimodal interaction	11.025%	2
Immersion features	Seat massage	0.708%	20
	Large angle seats	1.564%	17
	Ambient lighting	3.690%	10
	Smart audio	7.063%	5
	Separate sound zones	1.916%	15
	Active noise cancellation	0.708%	21
Security monitoring functions	Sentinel mode	4.934%	8
Security monitoring functions	Camera-monitor system (CMS)	10.017%	4
Cell phone-vehicle interconnection function	Mobile phone and car interconnection	1.211%	18
	Cross-screen connectivity between cell phone and vehicle systems	0.488%	22
	Bluetooth connection	4.570%	9
	WiFi connection	2.601%	13

Table 6 Sub-function hierarchy ranking results of mainstream IC function categories

Twelve datasets that did not meet the consistency criteria were excluded to ensure data reliability and validity. The CR of the final decision matrix was 0.084 (< 0.10), which confirming that the matrix satisfied the required consistency threshold in the AHP.

Hierarchy model of mainstream IC functions

As shown in Fig. 3, the mainstream IC functions of the in-vehicle system in the first hierarchy were organized into five function groups in the second hierarchy. Ranked in descending order of importance, these groups were: intelligent interconnection functions, interactive experience functions, immersion features, security monitoring functions, and cell phone–vehicle interconnection functions. The third hierarchy comprised detailed sub-functions corresponding to each of the five categories, with their rankings determined based on weight values, as presented in Table 6.

Figure 3.

The hierarchy model of IC mainstream functions.

Discussions

A hierarchical model of in-vehicle IC functions was established by integrating a questionnaire survey, PCA, and AHP. The results revealed a prioritized ranking of the mainstream function categories in descending order of importance: intelligent interconnection, interactive experience, immersive experience, safety monitoring, and cell phone–vehicle interconnection.

First, the intelligent interconnection category was identified as the most important, with a weight of 0.4225. Within this category, the center control screen received the highest subfunction weight. These findings align with previous studies,^15,16,37 which also ranked intelligent interconnection as a top priority. This prominence may be attributed to the high market installation rate and user familiarity with the head unit, a subfunction within this category, which was rated at 15.73% in the present study. This finding aligns with the work of Chang,¹⁴ who emphasized the head unit as a key component of user interaction. However, not all subfunctions within the intelligent interconnection category received high rankings, despite the category's overall dominance. This discrepancy may reflect variations in the installation rates of specific subfunctions, which can reduce their perceived importance among users. Furthermore, the results differ from previous studies that identified voice assistance as the most preferred function.^13,18 This difference likely arises from the inclusion of market installation rates in this model, which reflect actual adoption levels, whereas prior studies primarily focused on consumer preferences in purchase decisions. Additionally, the findings of this study do not fully align with Chang,¹⁴ who reported a high ranking for the camera–monitor system functions. This inconsistency may result from methodological differences, as this methodology employed targeted analytical methods tailored to the data sources at each research stage. Overall, these inconsistencies appear to stem from integration of practical market installation rates into the model. These rates reflect actual user preferences, combining subjective evaluations and voluntary adoption levels, and align with the theoretical framework of the technology acceptance model (TAM) proposed by Venkatesh and Davis.³⁸ Incorporating real-world data enhances the practical relevance of the model, providing a more grounded understanding of user behavior and function adoption than approaches based solely on perceived importance or theoretical assumptions.

Second, the sentinel mode within the safety-monitoring functions category exhibited high user demand but a relatively low installation rate (20%). During the market research phase, some consumers expressed resistance or aversion to the sentinel mode, citing concerns about reliability. Despite these reservations, overall demand for the sentinel mode remained substantial, likely driven by the increasing societal emphasis on safety, shaped by evolving socio-environmental factors. This finding aligns with,¹⁷ which noted that safety features strongly influence consumer adoption of new energy vehicles. Another possible explanation for the low adoption by manufacturers is cost constraints or technical challenges, as suggested by TechInsights.³⁹ These results are consistent with the unified theory of acceptance and use of technology (UTAUT).⁴⁰

Third, the cell phone–vehicle interconnection category ranked lowest in user demand despite prominent marketing in the automotive industry. This discrepancy may reflect users’ strong reliance on standalone mobile devices and their reluctance to invest in additional in-vehicle functionalities, as discussed by Mehler, Kidd.⁸ This finding is consistent with theories in human-machine interaction, which indicate that user behavior patterns significantly influence the design and adoption of in-vehicle cockpit features.³

Despite minor constraints from specific research contexts, the implications of this study demonstrate notable generalizability. Its core findings regarding the prioritization of in-vehicle intelligent cockpit functions, the integration framework of user preferences and market installation rates, and the application logic of technology acceptance theories (TAM / UTAUT) in in-vehicle systems are widely transferable to scenarios with similar automotive market landscapes, user groups, and technology penetration levels. These insights can provide robust guidance for in-vehicle intelligent cockpit function design, market positioning, and related technology acceptance research, and are also applicable to dynamic evaluations of in-vehicle function priorities amid technological evolution.

Limitations and future research

This study has several limitations that should be acknowledged. First, it focused on the Chinese market, and the findings require further validation for broader, global applicability. Specifically, context-specific installation rate data and local user behavior need to be collected to better reflect the technology acceptance patterns in different regions. Questionnaire participants should also be regionally representative to accurately capture local behavioral preferences. Second, the sample size and composition may not have been sufficient to support personalized market segmentation.

Future research could explore the derivation of user-specific features and the construction of personalized IC functional hierarchies, particularly in complex and dynamic driving environments, by employing a whole-to-part topological framework. Finally, this study primarily conducted a cross-sectional comparison of current mainstream IC features in the market. Future research could collect longitudinal data across multiple time periods to capture evolving user behaviors and needs in response to technological advancements and market changes.

Conclusions

This paper proposes a mainstream functional hierarchy model for intelligent cockpits (ICs) incorporating practical constraints such as functional installation rates. The model identifies user preferences for mainstream functions to support tailored function design/layout and analyze potential user needs. Key findings highlight intelligent interconnection as the most prominent function group with top development priority, and the central control screen as its core feature. These insights enhance driving experience, user engagement/loyalty, and automotive R&D efficiency, while providing theoretical guidance for optimizing IC function layout and targeted development.

Supplemental Material

sj-docx-1-wor-10.1177_10519815261434561 - Supplemental material for Hierarchical modeling of mainstream functional requirements for intelligent cockpits

Supplemental material, sj-docx-1-wor-10.1177_10519815261434561 for Hierarchical modeling of mainstream functional requirements for intelligent cockpits by Zhang Lu, Wang Xin, Xie Hui, Li Xiangrong, Yumitijiang Tohti, Zhao Fangyue and Yan Yulu in WORK

Footnotes

Acknowledgments

We thank all participants for supporting the experiments. Also, we would like to thank Editage (www.editage.cn</objidref>) for English language editing.

Ethical approval

This study was approved by the Ethics Committee of Tianjin University (No. TJUE-2024-139).

Informed consent

All participants were informed about the character of the study and the alignment with ethical and data privacy standards.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Supplemental material

Supplemental material for this article is available online.

References

Niewiadomski

. Incorporating model-based user requirements in product development. USA: University of Michigan-Dearborn, 2021.

Holtta-Otto

. American Society of Mechanical Engineers. Does empathizing with users contribute to better need finding? In Proceedings of ASME 2022 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, IDETC-CIE2022, Vol. 6 (pp. 1–8) [C]. St. Louis, MO, USA, August 14–17, 2022. 2022.

Dudziak

Kuranc

Zając

, et al. Smart solutions in car dashboard interfaces as a response to needs of drivers and their assessment. Case Stud Transp Policy 2024; 16: 101194.

Zhou

Jiao

Linsey

. Latent customer needs elicitation by use case analogical reasoning from sentiment analysis of online product reviews. J Mech Des 2015; 137: 071401.

KPMG. Intelligent Cockpit White Paper [Online]. Available: https://www.kpmg.com/cn/en/home/publications/2023/intelligent-cockpit-white-paper.html. [Last accessed: 2024-10-05]. 2023.

Eversdijk

de Vos

FJC

van Zee

AAT

, et al.

Does a reclined backrest with less legroom meet the same comfort as a fixed backrest with 80 mm more leg room?

Work 2024; 77: 103–113.

Lamble

Laakso

Summala

. Detection thresholds in car following situations and peripheral vision: implications for positioning of visually demanding in-car displays. Ergonomics 2010; 53: 807–815.

Mehler

Kidd

Reimer

, et al. Multi-modal assessment of on-road demand of voice and manual phone calling and voice navigation entry across two embedded vehicle systems. Ergonomics 2016; 59: 344–367.

Perrier

MJR

Louw

Carsten

OMJ

. Usability testing of three visual HMIs for assisted driving: how design impacts driver distraction and mental models. Ergonomics 2023; 66: 1142–1163.

10.

Di Ao

PKW

Huang

Mei

, et al. Analysis of co-relation between objective measurement and subjective assessment for dynamic comfort of vehicles. Int J Automot Technol 2020; 21: 1553–1567.

11.

Zhao

Yang

. Prediction models of overall thermal sensation and comfort in vehicle cabin based on field experiments. Int J Automot Technol 2024; 26: 269–282.

12.

Liao

Liu

Yang

, et al. An integrated assessment model of automobile smart cabin comfort based on weight optimization. Int J Automot Technol 2025; 26: 1709–1729.

13.

Dong

. Research on the influence of smart cockpit on the purchase intention of automobile consumers. J Shanghai Univ Finance Econ (in Chinese) 2021; 23: 112–125.

14.

Chang

. Research on intelligent function preference of new energy vehicles based on correspondence analysis. Automot Dig (in Chinese) 2024; 42: 28–35.

15.

Mihale-Wilson

Zibuschka

Hinz

. User preferences and willingness to pay for in-vehicle assistance. Electron Mark 2019; 29: 37–53.

16.

Luo

Song

Zhou

, et al. An analysis framework to reveal automobile users’ preferences from online user-generated content. Sustainability 2023; 15: 13336.

17.

Zhou

Adel

, et al. Examination of the adoption intention of new energy vehicles from the perspective of functional attributes and media richness. Heliyon 2024; 10: e25897.

18.

Lai

Lin

Zou

, et al. Kansei engineering for the intelligent connected vehicle functions: an online and offline data mining approach. Adv Eng Inform 2024; 61: 102467.

19.

Zhu

Sun

Zhao

, et al. How to rate the vehicle cockpit seat usage experience? A computational method based on embodied cognition theory. WORK: A J Prev Assess Rehabil 2025; 60: 457–470.

20.

Wen

Zhang

Zhan

, et al. The 2024 revision of the declaration of Helsinki: a modern ethical framework for medical research. Postgrad Med J 2025; 101: 371–382.

21.

Emary

Stuber

Mbuagbaw

, et al. Quality of reporting in chiropractic mixed methods research: a methodological review protocol. Chiropr Man Therap 2021; 29: 35.

22.

Bevan

NSR

. ISO 20282: Is a practical standard for the usability of consumer products possible? In Design, User Experience, and Usability (Vol. 6769, pp. 119-127). 2011.

23.

Automotive News. [Online]. https://www.autonews.com/. [Last accessed date: 2024-12-19]. 2024.

24.

Gasgoo Research. [Online]. https://autonews.gasgoo.com/about.aspx. [Last accessed date: 2024-12-19]. 2024.

25.

China Passenger Car Association. [Online]. https://www.cpcaauto.com/. [Last accessed date: 2025-12-19]. 2024.

26.

PCauto Club. [Online]. https://www.pcauto.com.cn/. [Last accessed date: 2025-12-19]. 2024.

27.

Vibhute

Gutierrez

Radivojevic

, et al. Multimodal web agents for automated (dark) web navigation. In Proceedings of the 17th International Conference on Agents and Artificial Intelligence (pp. 437-444). 2025.

28.

Palinkas

Horwitz

Green

, et al. Purposeful sampling for qualitative data collection and analysis in mixed method implementation research. Admin Policy Ment Health Ment Health Serv Res 2015; 42: 533–544.

29.

Zhu

Oberman

Aletta

. Defining acoustical heritage: a qualitative approach based on expert interviews. Appl Acoust 2024; 216: 109754.

30.

Dicle

Aletta

Kang

. A framework for developing adaptive acoustic comfort: insights from expert interviews. Appl Acoust 2025; 235: 110642.

31.

Nath

Islam

AMK

Majumder

, et al. Assessment of knowledge on human mpox virus among general population in Bangladesh. Discov Public Health 2025; 22: 1–12.

32.

Kai

. Application test of non-response Likert scale in tourism research. Tourism Trib. (in Chinese) 2011; 26: 29–35.

33.

CAAM. November 2024 CCAA Information Session [Online]. China Association of Automobile Manufacturers. Available: http://www.caam.org.cn/chn/4/cate_154/con_5236603.html. [Last accessed: 2025-02-20]. 2024.

34.

Hair

Risher

Sarstedt

, et al. When to use and how to report the results of PLS-SEM. Eur Bus Rev 2019; 31: 2–24.

35.

Tabachnick

Fidell

. Using multivariate statistics (6th ed.) [M]. USA: Pearson, 2018.

36.

Saaty

. A scaling method for priorities in hierarchical structures. J Math Psychol 1977; 15: 281–300.

37.

Yang

, et al. User requirements analysis of new energy vehicles based on improved Kano model. Energy 2024; 309: 133134.

38.

Venkatesh

Davis

. A theoretical extension of the technology acceptance model: four longitudinal field studies. Manag Sci 2000; 46: 186–204.

39.

TechInsights. Automotive system demand forecast [Online]. Available: https://www.199it.com/archives/1632938.html#google_vignette. [Last accessed: 2024-10-25]. 2023.

40.

Venkatesh

Bala

. Technology acceptance model 3 and a research agenda on interventions. Decis Sci 2008; 39: 273–315.

41.

. Intelligent Cockpit spread quickly, the penetration rate will exceed 70% this year [Online]. Gasgoo Automotive. Available: https://auto.gasgoo.com/news/202404/9I7038821. [Last accessed: 2025-02-01]. 2024.

42.

Kuien. Changes in the loading volume and penetration rate of the co-pilot screen | Gasgoo Automotive Configuration Database [Online]. Available: .https://auto.gasgoo.com/news/202312/6I70373222C601.shtml. [Last accessed: 2024-12-01]. 2023.

43.

IDC. China Passenger Car Market Data Tracking Report [Online]. International Data Corporation. Available: https://www.idc.com/getdoc.jsp?containerId=prCHC51965424. [Last accessed: 2025-02-05]. 2024.

44.

Institute, A. R. Where is the “intelligence” of China's smart cars? 2024 China Smart Electric Vehicle User Insights [Online]. Autohome Research Institute. Available: https://www.163.com/dy/article/JE5JED190527CRVH.html. [Last accessed: 2025-01-20]. 2024.

45.

CATARC. China Automotive Technology & Research Center Co. Ltd. [Online]. Available: https://www.catarc.ac.cn. [Last accessed: 2025-01-15]. 2024.

46.

GGII. UWB pre-installed has surpassed the milestone of 300,000 vehicles! The new round of competition for digital car keys is going on [Online]. Available: https://www.eet-china.com/mp/a297135.html. [Last accessed: 2024-11-25]. 2024f.

47.

Cloud

. Blue Book of Automotive Industry Analysis for the First Half of 2023 [Online]. Jiemian News. Available: https://www.jiemian.com/article/9823844.html. [Last accessed: 2024-11-05]. 2023.

48.

GGII. Data debut! The penetration rate of 360 panoramic front-loading is approaching 50%, and multiple benefits drive “new” increments [Online]. Baidu Baijiahao. Available: https://baijiahao.baidu.com/s?id=1809679932826652559. [Last accessed: 2024-12-20]. 2024c.

49.

Shujubang. 2023 Global and China Electronic Rearview Mirror Industry Research Report [Online]. Available: http://www.pday.com.cn/Htmls/Report/202306/24540549.html. [Last accessed: 2024-11-10]. 2023a.

50.

Shujubang. Monthly monitoring report on China's automotive intelligent cockpit technology and data trends 4 [Online]. Available: http://www.pday.com.cn/Htmls/Report/202406/24543635.html. [Last accessed: 2025-01-30]. 2023b.

51.

Shujubang. 2024 China Passenger Car Mobile Phone Wireless Charging Research Report [Online]. Available: http://www.pday.com.cn/Htmls/Report/202408/24544645.html. [Last accessed: 2025-02-10]. 2024.

52.

Autohome

. Keeping up with intelligence, expanding micromotors, ADAS cleaning and thermal management [Online]. Huaan Securities. Available: https://www.hangyan.co/charts/3321257488656893283. [Last accessed: 2025-03-10]. 2024.

53.

Shujubang. The installation rate of intelligent connected vehicles is approaching 30%, and car companies are accelerating the exploration of app store models [Online]. Available: https://zhuanlan.zhihu.com/p/384680940. [Last accessed: 2024-09-15]. 2021.

54.

Fient

. The “future seat” driven by intelligence, the consumption upgrade of the 100 billion track [Online]. Sina Auto. Available: http://k.sina.com.cn/article_2972154552_b1277ab8020013igs.html. [Last accessed: 2024-10-12. 2023.

55.

GGII. 2024 Global and China Automotive Comfort System Research Report [Online]. Available: http://www.pday.com.cn/Htmls/Report/202402/24543584.html. [Last accessed: 2025-01-08]. 2024b.

56.

GGII. The standard configuration rate of seat massage function is approaching the 10% mark, and independent brands/new forces are leading [Online]. EET China. Available: https://www.eet-china.com/mp/a346392.html. [Last accessed: 2024-11-30]. 2024e.

57.

GGII. Passenger car seat comfort system, seat memory function, front installation, standard model brand TOP 10 in Chinese market [Online]. EEO.com. Available: https://www.eeo.com.cn/2024/1010/690619.shtml. [Last accessed: 2024-12-15]. 2024d.

58.

GGII. 2023-2024 Global and China Automotive Lighting System Research Report [Online]. Available: http://www.pday.com.cn/Htmls/Report/202403/24543595.html. [Last accessed: 2025-01-08]. 2024a.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB