Project/Area Number |
17K00737
|
Research Category |
Grant-in-Aid for Scientific Research (C)
|
Allocation Type | Multi-year Fund |
Section | 一般 |
Research Field |
Design science
|
Research Institution | Toyota Technological Institute |
Principal Investigator |
|
Project Period (FY) |
2017-04-01 – 2021-03-31
|
Project Status |
Completed (Fiscal Year 2020)
|
Budget Amount *help |
¥4,680,000 (Direct Cost: ¥3,600,000、Indirect Cost: ¥1,080,000)
Fiscal Year 2019: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2018: ¥2,470,000 (Direct Cost: ¥1,900,000、Indirect Cost: ¥570,000)
Fiscal Year 2017: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
|
Keywords | 感性工学 / 意匠設計 / ロバスト設計 / 多目的最適化 / クラスタリング / 深層学習 / タグチメソッド / 人工ニューラルネットワーク / 畳み込みニューラルネットワーク / 敵対的生成ネットワーク / 応答曲面法 / 設計工学 / ロバスト最適化 |
Outline of Final Research Achievements |
In this study, a robust optimal design method based on fuzzy clustering, the Taguchi method, and a multi-objective genetic algorithm was proposed to generate product aesthetics that satisfy all customers even if their Kansei is diverse. In the case study, the proposed method was applied to a design of a car front face for 100 subjects. The results shows that the proposed method can generate the design that was highly preferred by all customers while minimizing the variation of preference among customers. The applicability of deep learning in Kansei engineering was also investigated, and it was shown that deep learning can learn and infer the relationships between product aesthetics and customer preferences, and can analyze the reasons for customer preferences.
|
Academic Significance and Societal Importance of the Research Achievements |
提案手法を用いることで,デザイナーの知識・経験に基づくのではなく,顧客の声(アンケート)に基づく製品意匠設計が可能になった.また,提案手法は,顧客の感性のばらつきの影響を低減することができるため,工業製品のように多数の顧客を対象とする製品にも適用可能な点が新しい. 感性工学における深層学習の適用については,その適用可能性を示した点に加えて,大量のアンケートに回答しなければならないという顧客の負担や,選好や印象の評価を学習データにすることによって生じる学習データの不確実性,不正確性など,今後考慮すべき課題を明らかにした点が新しい.
|