Integrating Usability in Automotive Navigation User Interface Design via Kansei Engineering

Balancing user requirements as well as usability requirements can be an exacting task for designers. Automotive navigation user interfaces are of no exception. The aim of this study was to integrate the concept of usability in the design process for automotive navigation user interfaces. A modified Kansei Engineering approach was used in order to define the concept of usability for automotive navigation user interface. Both regular and professional Malaysian drivers participated in the Kansei evaluation experiment (n=118). Results were analyzed with Principal Components Analysis (PCA) as well as Partial Least Squares (PLS) method. A finalized list of design specifications was developed and the design specifications were then used to develop a new automotive navigation interface design.


Introduction
Mobile connectivity has become part and parcel of today's society.The demand of today's society is such that the need for information exchange between individuals across the globe must occur in an instant.Mobile phones have transformed themselves into smart phones with various functions for leisure and business productivity.Similarly, the public demands that the vehicles are also equipped with features similar to their smart phones.
Hence, the rapid development of automotive navigation user interface has brought forth the concern for the development of safe and usable automotive user interfaces systems for the end users.Usability and aesthetics need to go hand in hand even though aesthetics and actual usability may not be related at all (M.Mohamed, 2011).
Compared to the mobile phones market, the automotive user interface systems need to cater the primary concern of the driver, which is safety (Mohamed et al, 2015).In order for users to enjoy features of their automotive navigation user interface, the design of automotive navigation user interface system needs to promote and enhance the safety of drivers.However technological advancement of automotive navigation user interface systems has become so rapid that it is a tall order for manufacturers to ensure seamless interactions between automotive navigation user interface and drivers.Car dashboard designs have changed substantially from being cluttered with buttons to a single touch screen system with multiple functions.Usability and ergonomics have started to come into the picture where the interaction between drivers and dashboard is mentioned.(M.Mohamed & Mustafa, 2014) A popular concept which is often discussed in interface design is the concept of mental models.A mental model is a representation of a certain process or the surrounding environment in a human mind (Wickens, 1992).Mental models influence how a person reacts to their surrounding environment, and how a person perceives their environment.In the case of developing safe and usable products, the products should match a person's mental model to ensure a positive user experience.Users come with their own mental models; and the designers often have their own mental models when developing products.A mismatch often occurs with designer's mental models and the user's mental model; thereby causing problems when users interact with products (Lim, Pangam, Periyasami, & Aneja, 2006).
With the popularity of automotive navigation systems in cars, designing usability into automotive navigation systems is a real challenge for designers.Designers and users often have different conceptualizations of what usability is.This phenomenon was illustrated by Norman & Ortony (2003) by the concept of affordances between users and designers.Designers may try to influence the user's emotion through their product design (visceral and behavioral aspects), but users may perceive the design differently.Currently, there is little research bridging the gap between the users' and designers' mental models for automotive navigation interface design.Researchers such as Chang (2010), Knapp (2007), Winter et al (2009) looked at the differences that arose from different user backgrounds when interacting with a user interface.Different users tend to have different mental models; therefore there is an acute need to accommodate different mental models' of users when designing products such as automotive navigation interface design.Researchers and designers do understand that differences exist, but what can be done in order to address the issue on hand?Designers need a practical method or tool to integrate safety and usability during product design process.The design stage of a product is the most crucial; this is where the changes can be made to accommodate various user requirements.
Various ways have been proposed in order to improve the safety and usability of automotive user interfaces during the design process.Popular methods for integrating safety and usability have been primarily dominated by expert-led techniques such as rules and guidelines related to automotive user interfaces (Alliance Guidelines, 2002;Bhise, 2012;Green, Levison, Paelke, & Serafin, 1993).In expert-led techniques, user inputs during the design process are not considered at all.Expert-led techniques do not take into account how the users might actually perceive the product or system in use.A user-led technique such as Kansei Engineering (KE) for example, puts the user at the forefront where their perceptions of the product to be designed are taken into account during the design process.The word "Kansei" means total emotions, those feelings that are elicited by products or services (Schütte, 2005).KE is among the earliest technique of customer oriented design method developed in Japan by Mitsuo Nagamachi.One of the successful examples of KE was the birth of Mazda Miata, which fully utilized KE technique.The Mazda Miata became one of the best selling cars in history due to the implementation of KE in its design process, as all the relevant aspects for customer's emotional needs such as structural, functional and aesthetics were considered (Nagamachi & Lokman, 2010).As a precursor to this study, the concept of usability related to automotive navigation user interface research was carefully defined by Mohamed et al (2015a), where it was narrowed down to eleven variables as shown in Figure 3.

Method
A modified Kansei Engineering approach was utilized in this study.Instead of focusing on the emotional and aesthetics, safety and usability were the primary focus, as proven by an earlier research on car center stack designs by Mohamed et al (2015b).Figure 1

Design Elements Analysis
In order to develop Kansei equations, each of the design samples needs to be categorized according to their specific design elements.The more specific the design element the more accurate the Kansei equations will be.A detailed list of the classification is shown in Table 1.According to Mohd Lokman ( 2009) the selection of Kansei design samples needs to fulfill these criteria: 1) For each design sample, only ONE element from each design group can be selected, for example "Road Name Location" for a design sample must only have one value checked, such as "Bottom Center".
2) Only one design sample is selected if two or more samples have exactly the same design elements value.
3) Two or more design samples will be selected if they have some common design elements.

Selection of Design Samples
A proper selection of automotive navigation interface design samples needs to be accomplished before the Kansei measurement.Images of the interface design specimens followed the typical automotive navigation screen dimensions of 4 to 5 inches with a minimum resolution of 800 x 600 pixels for clarity.
Briefly, all the automotive navigation interface design samples selected needs to feature a complex road geometry system, urban roads, and only designs featuring English and Malay language were selected.The complex road geometry and the urban roads were chosen as most drivers have difficulty in navigating using automotive navigation if the road design features multiple junctions and curves, as stated by Nowakowski et al (2003).Automotive navigation interface design samples were restricted to the ones featuring Malay and English language as the participants in the study were all Malaysians.A pilot study done by the authors Mohamed et al. (2015) narrowed down the design samples selection to fourteen ( 14) design samples to be chosen for the Kansei survey.

Refinement of Kansei Words related to Safety and Usability
In this study, since the primary focus is on Kansei words related to safety and usability, Kansei words selected are related to safety, ergonomics and usability.The Federal Motor Vehicle Safety Standards, as well as a checklist developed by Bhise (2012) were used as the main reference for obtaining Kansei words related to safety, ergonomics and usability.Eleven Kansei words are selected as shown in Figure 2. The Kansei words are as follows: Each one of these Kansei words was accompanied with a Malay/English description so the words would not be misconstrued by the participants.

Kansei Measurement and Analysis
In the Kansei evaluation experiment, two types of Malaysian drivers participated, regular and professional.A total of 20 professional drivers, 53 male drivers and 45 female drivers participated during the evaluation experiment, while 69% of the participants were between 18-25 years old, 20% were between 26-35 years old, 10% were from 36-45 years old and 1% was between 46 -50 years old.Each of the participants had to evaluate 14 automotive navigation interface design samples using the list of Kansei words (listed in Figure 3).A 7 point semantic differential scale was used during the evaluation experiment.Each Kansei word was described as such in Figure 4 below for better clarity and understanding for the participants.The results were then analyzed using Factor Analysis and Principal Component Analysis.Lastly, several Kansei equations, linking design elements and Kansei words were developed for the final outcome of this research.
Figure 3. Kansei words with accompanying descriptions

Results and Discussions
A Principal Components Analysis (PCA) was conducted on the data, and only one component was revealed by the analysis, which is consistent with what the researchers are trying to focus in this study (i.e usability).The initial eigenvalues showed that one factor explained 96.892% of the variance.All of the variables contained in that component showed high loadings (minimum of 0.977 and above).The Cronbach's Alpha value is 0.980, indicating very high data reliability.
Kaiser Meyer Olkin values for sampling adequacy was 0.756, and the Bartlett's Test for Sphericity was significant at (χ2 (55) = 344.866p < .05)(Table 1).The diagonals of the anti-image correlation matrix were between 0.698 to 0.839, supporting the inclusion of each item in the PCA analysis.Details of the PCA analysis can be seen in Table 2.

Relationship between Usability and Automotive Navigation Interface Design Elements
A partial least squares (PLS) analysis was performed in order to establish the connection between the automotive navigation interface design elements and usability.The evaluation ratings for each Kansei word listed in Figure 2 were treated as dependent variables, and the design elements classification as independent variables.The results from the PLS analysis would show the relative weightings for each design elements with respect to a particular Kansei word.A high positive weighting indicates high degree of correlation with the Kansei word, and a negative weighting indicates otherwise.Figure 4 shows a sample of the PLS analysis results.For each of the Kansei words listed, all the highest standardized coefficient values were selected as the relevant design elements specific for that particular Kansei word.A sample of the analysis is shown in Figure 4.For example, as illustrated in Figure 4, in order for an automotive navigation interface to be "Operable", the road name color and location are not required.The highlighted route should be purple in color, and the map background color should be green.When all of the relevant design elements for each Kansei word were obtained, a finalized list of design specifications for a new automotive navigation interface was made, as shown in Figure 5.The finalized list contains the common design elements across all the 11 Kansei words related to usability.From the finalized list of the design specifications, an automotive navigation interface concept which is centered on usability was developed by a graphics designer.
Figure 1.Simplified Kansei Engineering implementation framework

Figure 2 .
Figure 2. List of Kansei Words

Figure 4 .
Figure 4. Sample of PLS Analysis Results

Figure 5 .
Figure 5. Finalized list of design specifications for usable automotive navigation interface

Table 1 .
Design elements classification

Table 2 .
PCA Analysis Results