Human-centric Product Design with Kansei Engineering and Artificial Intelligence
An introduction to Kansei Engineering Process, followed by the advancements through AI and Big Data to capture consumer affection and emotions.
“The designer does not begin with some preconceived idea. Rather, the idea is the result of careful study and observation, and the design a product of that idea.” -by Paul Rand
The ultimate goal of most of the products/solutions is to provide the best utilization and satisfaction to end-users/ customers. The satisfaction and the likeness of customers are most likely to be decided by human emotion and affection in mind. The concept of Kansei Engineering (founded by Mitsuo Nagamachi, 1974) aims to capture and bring customers’ psychological feelings into the design of products or social systems. The psychological feelings in this sense include concepts such as Want, Need, Aesthetic sensation (beautiful, elegant, etc.), good taste, etc. Kansei Engineering begins with an observation of a customers’ behavior, grasps their emotion and affection using psychological scaling, and analyzes the emotions using several statistical analyses to lead design specifications of products or systems.
First, let’s try to have an intuition of the concept of Kansei. The following image shows two pieces of art. One is called Lumumba and the other is called Takete. Just by looking at the image, can you name them as either Lumumba or Takete?
Most of you would have named the jagged art to the left (A) as Takete and the curvy art to the right (B) as Lumumba. Similarly, Ramachandran et al. [1] conducted an experiment with college students and undergraduates, which resulted in 98% selecting the left as Taketa and the right art as Lumumba. This suggests that the human brain is able to extract abstract properties from shapes and sounds, and they are most likely associated with psychological feelings.
Above is a simple experiment of understanding the psychological feelings of a human, but in the real-world the subject (domain) of concern can be a new mobile phone, sports item, website or even an AI product, instead of an art. As such, Kansei Engineering is a methodology (or an art) of capturing such emotions and affections of people (specifically targeted consumers) and translate into product designs and solutions.
Traditional Kansei Engineering Process
The Kansei Engineering Process has been utilized for the development of a number of product domains such as household items, gardening tools, vehicles, garments from its inception. The traditional view of the Kansei process is depicted in the following image.
In principle, we have an affective (emotional) response and a rational (cognitive) response for every sensory. That is, when a sensory input is received, the human will construct some type of a conscious that leads to affection and emotion. Sometimes, it could be difficult to provide an explanation as to why the emotion was generated because similar to the Taketa and Lumumba example, it is very subjective and implicit. But most of the time, consumers will decide on products based on the first impression, thus, having an understanding of the consumers’ thought process is quite important. On the other hand, the human will construct a rational understanding of the sensory perception as well. That is a quest for a rational response from the human on to what he/she thinks of the product. It is not necessary that both the responses are alike, but in the design of the product, Kansei Engineering process consider both the responses into account to drive the final product design.
Thereby, the Kansei Engineering Process flow can be understood based on the following flow diagram presented. In the beginning, the domain is defined, which is the context that describes the overall idea behind the product. This includes understanding and defining the intended target group and user type, market-niche, and the product group in question.
The domain is then described in two different aspects. One is semantic space and the other is the properties/specifications of the product. Semantic space can be described in a certain vector space defined by semantic expressions (words). This can be done by collecting a large vocabulary of words that describe the domain from a semantic view. These words are named Kansei (emotional) words. For example, to describe a mobile phone the words like cool, elegant, nice, handy can be selected.
The space of application can be defined by understanding the properties that are really important for the users. This could be key features and selects product properties for further evaluation. Such properties can be identified from different sources such as existing products, customer suggestions, possible technical solutions and design concepts.
The next step is unique in Kansei Engineering, which is the synthesize of the two spaces. Techniques such as Category Classification Method, Psychological Scaling Method, multi-variable statistical analysis and Kansei Rough Set Model have been used to conduct this step in recent literature. Taking the same example of a mobile phone design, this step will analyze what application specifications can drive the semantic space indicators such as cool, handy, beautiful and vise-versa. Few of the widely used techniques for synthesizing application and emotion spaces are briefly introduced below. Further details can be found by this research article by Prof. Mitsuo Nagamachi, who is the founder of Kansei Engineering Process [2].
(1) Category Classification Method: First decide the new product domain: (e.g., online market place, wireless accelerometer). Imagine the product and write the customers’ emotion, for instance, imagine what kinds of property the customers need, and then write down them on a piece of paper. Then, create a tree structure using the cards, like the first category, second category and so on. A “Fish-born diagram” can be used in this stage.
(2) Psychological Scaling Method: After collecting Kansei (emotional) words, you create a Likart-type scale using those Kansei words (or SD scale). In this case, the 5- or 7-scale are very popular. In this method, we usually look at the real samples of the product and check the feeling on the scale. Next, analyze these records using the multi-variable statistical methods.
(3) Analysis using the multi-variable statistical method: Calculate Factor Analysis to get a similar meaning group as a factor structure and delete the meaningless Kansei words from the research. Principal Component Analysis, data mining method, and others are useful. The JMP (SAS) analysis is most useful to find the design items from customer emotion.
(4) Kansei Rough Set Model: Rough Set Model is a unique mathematical model developed by Professor Zdzislaw Pawlak in Poland and it is able to make clear the relation between hidden emotion and design items.
Following the synthesize step, Validation and Model building steps are conducted in order to check if the prediction model is reliable, realistic and suitable. In case of prediction model failure, it is necessary to update the Space of Properties and the Semantic Space, and consequently refine the model.
Products developed based on Kansei Engineering
A number of large companies have utilized this Kansei Engineering process, specifically in Japan and Sweden. Companies such as Toyota, Honda, Mazda, Panasonic, Sanyo, Samsung, Sony, Komatsu, Yamaha are few such world-recognized companies that use the Kansei process. Some of such products are, New refrigerator design by Sharp (1979), MX5 passenger car by Mazda (1987), Sonata 2 by Hyundai (1995), Boeing 787 interior design (2008) and Urban Rider Jeans by Vf Lee (2014).
The Mazda MX5 (1987) was an interesting case study — ideation of the CEO of Mazda for a concept car that would be a sport-type passenger car for the young generation.
As detailed in the original article by Nagamachi et al. [2] a group of young drivers was selected as the user group. Two sets of drivers were taken. The first took place in the seat next to the young driver. A video was taken when the driver is on the wheels. The second was where researchers stand at the intersection and start a video camera, if any driver looks like a young driver. After this survey, all R&D members gather into a research room and write the driver’s emotion or action, each one on a small card looking at video show. This is called as the Category Classification Method. After then, they collected the cards on one group, if they are supposed as the same category. After that, R&D members make a tree structure using the cards like a fish-born diagram. Looking at the tree structure, the designer group linkage card meaning to design elements.
Augmenting Kansei Engineering Process with AI and Machine Learning
The traditional Kansei Engineering process was conducted using personalized experience samples which included 10–100 consumers. However, with the proliferation of Big Data, a vast volume of data is freely and publicly available today. For example, social media platforms such as Twitter, Facebook, Instagram contain millions of sentiments, emotions that can extract from the general public. In addition, product review websites (e.g., Online Product Reviews) and online discussion forums (e.g., Whirlpool Online Forum) also contains volumes of people’s opinions, expressions and reviews. Further to that, a large number of surveillance cameras setup at supermarkets can capture facial expressions, emotions of buyers of different products. Utilizing these volumes of big data will definitely enable to augment the Kansei Engineering process taking it to the next level. As such, we extend the traditional Kansei Engineering process architecture to be augmented by using AI and Big Data as following model architecture.
Consider the development of a new wearable smartwatch that can capture inertial measurements of a person. Rather than designing a prototype, we can simply design a Virtual Reality (VR) application for users to experiment with the product while capturing user behavior through video cameras — to capture expressions. In addition, the semantic space can be captured by online social media, product reviews and discussion forums for similar products to identify public perception. The application space can be captured similarly by such online media. After that, following through the Kansai Engineering Process the product design can be improved and realized.
Materializing the emotion extraction from online social media data, can be utilized in multiple domains as it provides an opportunity to understand the customer/stakeholders in a better way. Recently the Centre for Data Analytics and Cognition team developed the Patient-Reported Information Multidimensional Exploration (PRIME) framework that aims to identifying affective computing aspects in digital patient care [3]. In addition to social media data, a number of human activity recognition and emotion detection methods from image and video data are available which can be used to derive affective information regarding the user [4].
Below video shows Emotion Research LAB’s facial recognition software captures the emotions in real time of a consumer while testing a yogurt.
Final Thoughts
Despite the availability of a number of product design and developments methods , I’m fascinated by the Kansei Engineering Process that brings customers’ psychological feelings such as Want, Need, Aesthetic sensation (beautiful, elegant, etc.), good taste, into the design of products. The Kansei Engineering Process has been utilized for the development of a number of product domains such as household items, gardening tools, vehicles, garments from its inception, and being utilized by a number of world-leading industries such as Toyota, Mazda, Sony, Samsung, Panasonic, etc. In a nutshell, Kansei Engineering Process takes consumer emotion (semantic space) and cognitive understanding of application specification (application space) in to consideration in the design of products.
Traditional implementations of the Kansei Engineering process consider experience tests using a limited number of consumers (due to feasibility reasons), however, with the advent and proliferation of Big Data, the vast volume of data that is freely and publicly available today can be harnessed to generate a better, sound and sophisticated view of the semantic space and application space. Several tools and techniques are today available to maximum utilization of such Big Data platforms. Thereby, utilizing such tools and techniques in combination with Kansei Engineering process has potential to design consumer friendly products, systems and solutions that can reach the mass market.
References
[1] Ramachandran, Vilayanur S., and Edward M. Hubbard. “Hearing colors, tasting shapes.” Scientific American 288.5 (2003): 52–59.
[2] Nagamachi, Mitsuo. “History of kansei engineering and application of artificial intelligence.” International Conference on Applied Human Factors and Ergonomics. Springer, Cham, 2017.
[3] Adikari, Achini, et al. “Can online support groups address psychological morbidity of cancer patients? An artificial intelligence based investigation of prostate cancer trajectories.” Plos one 15.3 (2020): e0229361.
[4] Nawaratne, Rashmika, et al. “Hierarchical two-stream growing self-organizing maps with transience for human activity recognition.” IEEE Transactions on Industrial Informatics (2019).
