INFERENCE OF LATENT FUNCTIONS IN VIRTUAL FIELD

Size: px

Start display at page:

Download "INFERENCE OF LATENT FUNCTIONS IN VIRTUAL FIELD"

Dylan Woods
5 years ago
Views:

The Fourth International Conference on Design Creativity (4th ICDC) Atlanta, GA, November 2 nd -4 th, 2016 INFERENCE OF LATENT FUNCTIONS IN VIRTUAL FIELD S. Fujii 1, K. Yamada 2 and T.

1 The Fourth International Conference on Design Creativity (4th ICDC) Atlanta, GA, November 2 nd -4 th, 2016 INFERENCE OF LATENT FUNCTIONS IN VIRTUAL FIELD S. Fujii 1, K. Yamada 2 and T. Taura 1,2 1 Department of Mechanical Engineering, Kobe University, Kobe, Japan 2 Organization for Advanced and Integrated Research, Kobe University, Kobe, Japan Abstract: Products may possess latent functions that might be eventually discovered by users. We propose a method of inferring the latent function of a product by using it in fields different from that intended by the designers. In particular, we use a virtual reality device to realise various fields of application for the product and experimentally investigate the manners in which the types and number of latent functions inferred correlate with the strength of the new product field relationships. The results show that more functions are inferred in fields that are strongly related to the product and that more novel latent functions are inferred in fields that are weakly related to the product. Keywords: latent function, virtual reality, field 1. Introduction User innovation is one of the principles of product innovation. According to the user innovation theory, product innovation is promoted not only by manufacturers but also by users (Hippel, 1976). For example, DJ turntables were created through user innovation (Faulkner & Runde, 2009): Turntables were originally designed as playback devices. In 1997, Theodore Livingston (stage name Grand Wizzard Theodore) invented scratching, a sound manipulation technique in the art of turntablism; along with the development of other such techniques, this contributed to turntables being designed as not only playback devices but also musical instruments. Thus, users discovering uses for products can lead to product innovation. New uses can be discussed using function terminologies as discovering hidden functions of the products. In the turntable example, the hidden function <make sounds as musical instrument> 1 was discovered by a user. Such a hidden function is termed a latent function (Yoshikawa, 1981). Latent functions are usually discovered through use of the product, as was the case with DJ turntables. However, latent functions are expected to be inferred before actual use because of these characteristics: 1 This paper uses < > to indicate specific functions; definite and indefinite articles may be omitted to simplify these phrases grammatically. 4th ICDC 1

2 Latent functions may endanger users and those around them. For example, although the original function of a kitchen knife is to cut foodstuffs, it possesses the latent function of accidentally or intentionally injuring the user or those around them. Ideally, product design should preclude the manifestation of such dangerous latent functions. Latent functions can enhance product value and initiate demands for new products. For example, an empty can may be used as a flowerpot, which may lead to the design of flowerpots that resemble cans. Thus, inferring latent functions may expand the variety of product uses, enhance product value, and spur the development of new products. Moreover, visualizing hidden risks can help prevent accidents (Taura, 2016). Two methods have been reported for inferring latent functions. The first method is based on word replacement (Mori et al., 2013). In functions that are described as a set of verbs and objects, the verb or the object can be replaced with others verbs and objects. For example, the latent function of using a digital camera as a memo can be inferred by replacing scenery in its function <to record scenery> with text. The second method employs analogies (Minami et al., 2013). This method is based on the idea that if two products have a similar structure, the function of one product could be a latent function of the other. For example, a desk and a footstool have similar structures; therefore, a desk can be used as a footstool when a person wants to reach something placed at a height. Yoshikawa (1981) defined a latent function as a new function that is manifested by a product in a different situation (field) from a situation (field) in which designers assumed it will be used ; as an extension of this idea, latent functions can be inferred by intentionally attempting to use the target product in a field that is different from the originally assumed field. Hence, in this study, we used a virtual reality (VR) device to realise different fields and experimentally investigated the types of latent functions that can be inferred from the new relationships created between the products and fields. 2. Strength of the product field relationship First, we introduce the concept of the strength of the relationship between products and the fields in which the products will be used. For example, the relationship between a paper cup and a picnic in which the paper cup will be used is strong, whereas that between a paper cup and a desert in which the paper cup will be used is weak. Many functions, such as <put garbage>, <catch insect>, and <take drinks>, can be easily identified in the picnic field, but these functions are largely inapplicable in the desert field. Nevertheless, some novel latent functions, such as <cover ears> to protect the ears from a sandstorm, may be inferred in the latter. Therefore, we hypothesize that more functions would be inferred in fields that are strongly related to the product, whereas more novel latent functions would be inferred in fields that are weakly related to the product. 3. Virtual fields The VR device π-cave, a variant of CAVE (Cruz-Neira et al., 1992), was used in this study. π-cave is composed of four upright screens (two in the front and one each on the sides) and two screens on the floor (Fig. 1), onto which a visual image can be projected. π-cave has a rectangular parallelepiped configuration with dimensions of 3 m 3 m 7.8 m. Users can see the stereoscopic visual image by using three-dimensional glasses equipped with a tracking device, which provides users the experience of walking through virtual space as they walk on the floor. In this study, we created virtual fields in which new uses of products were sought by projecting a panoramic image (180 panorama in the vertical (up down) direction and 360 panorama in the 2 4th ICDC

horizontal (left right) direction) and onto π-cave. This was realised by mapping a 360 panoramic image on the inside of the sphere and placing the camera in the centre of the sphere in the software.

Thus, because the virtual fields generated by π- CAVE are realistic, we expect novel latent functions to be easily inferred without the need for real fields. We used clothes pegs (Fig.

3 horizontal (left right) direction) and onto π-cave. This was realised by mapping a 360 panoramic image on the inside of the sphere and placing the camera in the centre of the sphere in the software. The camera direction can be changed using a controller. Thus, because the virtual fields generated by π- CAVE are realistic, we expect novel latent functions to be easily inferred without the need for real fields. We used clothes pegs (Fig. 2) as the experiment product because they are widely used; we used garden and kitchen as the experiment fields strongly related to clothes pegs, bike parking lot and space (universe) as fields weakly related to clothes pegs, and forest as the field that is neither strongly nor weakly related to clothes pegs (see Table 1 for the 360 panoramic images). In the experiment, we asked the participants to evaluate the strength of the relationship between the clothes pegs and the aforementioned fields by ranking the fields in the descending order of the relationship strength. Figure 1. External view of π-cave 2 Figure 2. Clothes pegs used in the experiment 2 π-cave Introduction YouTube: Accessed Jun th ICDC 3

4 Table 1. Virtual fields used in the experiment 360 panorama image Field Garden Kitchen 3 Bike parking lot Space 4 Forest Accessed Jan Accessed Jan stream/. Accessed Jan th ICDC

5 4. Experiment 4.1. Participants Seven male university students (mean age, 20 years) participated in the study. Overall, the participants provided 128 uses (functions) Experimental procedure Each participant sequentially participated in the experiment. Participants were allowed 3 min in each of the five fields and were asked to identify the uses (hereafter, functions) of clothes pegs relevant to each field. The order of the fields was randomized for each participant. The following instructions were provided to the participants: The virtual field projected in π-cave can be explored using the controller. The function must be described in the form <verb object>. If a function is relevant to multiple fields, the function must be described in all relevant fields. Multiple, different-sized clothes pegs can be used when identifying or creating functions. The clothes pegs can be freely modified as long as they can still be recognized as clothes pegs. Subsequently, the participants were shown the 360 panorama images displayed in Table 1 and were asked to intuitively rank the strength of the relationship between the clothes pegs and the five fields. 4.3 Analysis The three authors of this paper independently classified all identified functions into two types: ordinary and novel. Functions judged ordinary by at least two authors were considered ordinary for the subsequent analysis. Of the 128 functions identified by the participants, 46 were classified as ordinary, and the remaining were considered the novel latent functions in this study. Kendall s concordance coefficient for the classification revealed moderate concordance among the authors. First, all identified functions (ordinary and novel) were analysed on the basis of the strength of the relationship between the product and the fields. The number of functions identified in the field ranked as the most closely related to the product was summed. This summing process was repeated for the fields ranked as the second, third, fourth, and fifth most related to the product. Next, this analysis was repeated solely for the novel latent functions. 5. Results Table 2 shows the strengths of the relationships between the product and the fields as ranked by the participants. Table 2. Strengths of the product field relationships as ranked by the participants Kitchen Garden Bike Parking Lot Forest Space Participant 1 1st 2nd 3rd 4th 5th Participant 2 1st 2nd 3rd 4th 5th Participant 3 1st 3rd 2nd 4th 5th Participant 4 1st 2nd 3rd 4th 5th Participant 5 2nd 1st 3rd 5th 4th Participant 6 2nd 3rd 4th 1st 5th Participant 7 1st 3rd 2nd 4th 5th 4th ICDC 5

6 5.1. Correlation between the number of functions and the strength of the product field relationship Table 3 presents the number of functions identified for each rank of the strength of the product field relationship. The total number of functions negatively correlated significantly with the strength of the relationship (Spearman s rank correlation coefficient ρ =.900; p =.037) (Fig. 3), indicating that more functions could be inferred in fields strongly related to the product than in those weakly related. Table 3. Number of functions (ordinary and novel) for each rank of the product field relationship strength 1st 2nd 3rd 4th 5th Participant Participant Participant Participant Participant Participant Participant Total Figure 3. Correlation between the total number of functions and the strength of the product field relationship 5.2. Correlation between the number of novel latent functions and the strength of the product field relationship Table 4 presents the number of novel latent functions for each rank of the strength of the product field relationship. The total number of functions correlated significantly with the strength of the relationship (ρ =.900; p =.037), indicating that more novel latent functions could be inferred in fields weakly related to the product than in those strongly related. 6 4th ICDC

7 Table 4. Number of novel latent functions for each rank of the product field relationship strength 1st 2nd 3rd 4th 5th Participant Participant Participant Participant Participant Participant Participant Total Figure 4. Correlation between the total number of novel latent functions and the strength of the product field relationship 5.3. Examples of novel latent functions The following are examples novel latent functions identified by the participants. Note that our selection was rather subjective. Visualize differences in expiration dates of foodstuffs by attaching clothes pegs of different colours to them (kitchen) Create a drain spout by connecting several clothes pegs and hanging them from a roof (garden) Mark one s own bike by using clothes pegs with conspicuous colours (bike parking lot) Prevent one from getting lost by placing clothes pegs on paths that have been passed (forest) Create a fishing float by using clothes pegs (forest) Create a tent frame by connecting numerous clothes pegs (forest) Plug a hole in a spacesuit by using a clothes peg (space) Generate driving force by throwing clothes pegs (space) Fabricate a spaceship by using clothes pegs (space) 6. Conclusion Yoshikawa (1981) defined a latent function as a new function that is manifested by a product in a different situation (field) from a situation (field) in which designers assumed it will be used ; accordingly, we proposed a method of intentionally creating new relationships between products and fields by using a VR device. We introduced and validated the concept of the strength of the relationship between products 4th ICDC 7

8 and the fields in which the products will be used. The results showed that more functions could be inferred in fields strongly related to the product and that more novel latent functions could be inferred in fields weakly related to the product. Thus, the results confirm that the strength of the product field relationship is a feasible guideline for creating new product field relationships and that VR is effective in creating new product field relationships. In future studies, we plan to create new and unique fields by using VR to further investigate the inferring of novel latent functions. Acknowledgement We thank Mr Yuusuke Murayama for his kind cooperation. This work was supported by Kaken (15K12291). References Cruz-Neira, C., Sandin, D. J., DeFanti, T. A., Kenyon, R. V., & Hart, J. C. (1992). The CAVE: Audio visual experience automatic virtual environment. Magazine of Communications of the ACM, 35(6), Faulkner, P., & Runde, J. (2009). On the identity of technological objects and user innovations in function. Academy of Management Review, 34(3), Minami, K., Taura, T., & Tsumaya, A. (2013). A study on modeling of reasoning process considering latent function. Journal of the Japan Society of Mechanical Engineers (C), 79(803) (in Japanese) Mori, H., Taura, T., & Tsumaya, A. (2013, August). Method for inferring latent functions. In Proceedings of 19th International Conference on Engineering Design, ICED13 (pp ). Taura, T. (2016). Creative Design Engineering, Waltham: Elsevier. von Hippel, E. (1976). The dominant role of users in the scientific instrument innovation process. Research Policy, 5(3), Yoshikawa, H. (1981). General design theory and a CAD system. In: T. Sata & E. A. Warman (Eds.), Man-Machine Communication in CAD/CAM, 35 53, Amsterdam: North-Holland. 8 4th ICDC

Determining Latent Functions by Focusing on the Purposes of. Products

5th IADR 03 TKY Determining Latent Functions by Focusing on the Purposes of Products Yuki Murota*, Toshiharu Taura**, Akira Tsumaya*** *Kobe University, murota.yuki.mp3@is.naist.jp ** Kobe University,