虛擬性別：視覺提示對機器人性別化之影響

Virtual Gender: The Impact of Visual Cues on Robot Gender

洪唯慈(Wei-Tsz Hung)

指導教授 ： 游曉貞

國立臺中科技大學/設計學院/多媒體設計系碩士班/碩士(2020年)

摘要

在現代科技日新月異的時代中，有越多接近人類智能特質或是以擬人特質的數位科技產品，如：服務型機器人、虛擬助理…等。隨著性別化科技概念的興起，機器人與使用者的人機互動(Human-Robot Interaction，HRI)研究領域中，也開始重視人工智慧的性別化議題的，希望理解賦予人工智慧性別對使用者或HRI的影響。在此脈絡下，本研究著重於使用者如何理解機器人的虛擬性別來進行探索，特別是如何透過不同的外觀設計影響人們對機器人性別角色的評斷。本研究利用感性工學(Kansai Engineering)的方法進行研究：在分析目前市面上之機器人設計元素後，選出4種較影響機器人性別的設計項目，每個項目各有3個類目：裝飾色（藍色、粉紅色、無色彩）、底色（白色、淺灰色、深灰色）、配件（領帶、蝴蝶結、無配件）及髮長（無髮、長髮、短髮）。根據L9 (34)田口直交表將潛在的81種設計組合，簡化為 9 個機器人實驗條件，並以UBTECH的Lynx機器人為雛型，用影像編輯軟體繪製出實驗樣本。本研究直接採用「12題班氏性別角色量表」（BSRI-12)中6個陽剛形容詞及6個陰柔形容詞，加上1題性別主觀判斷之形容詞語對（男-女），共 14 個形容詞作為五點語意差異量表之感性語彙。透過社群平台在線上匿名施測，回收 280 份有效問卷，經數量化一類分析，本研究發現：(1) 機器人的外觀上的視覺提示確實會影響受測者對它們的性別角色（陽剛或陰柔）及性別（男或女）的感知；(2) 利用外觀設計作機器人的性別提示時，不同性別的受測者對機器人的性別感受（陽剛或陰柔、男或女）評價也有差異，各種視覺元素對男性或女性受測者的影響不同。本文最後提出對機器人性別化設計之建議，可作為未來在開發不同用途的機器人的參考依據。

關鍵字

人與機器人互動 ； 性別提示 ； 感性工學 ； 12題班氏性別角色量表

並列摘要

With the rising of “Gendered Innovations”, gendering social robots has become an issue in the field of Human Robot Interaction (HRI), and we want to find out the impacts of endowing artificial intelligence with genders. Under the premise, this study focuses on how users understand the virtual genders of robots; especially, how design elements of robot arouse people’s gender perception of robots by using the Kansei engineering method. This study uses the Kansei Engineering. First, collect and analyze the gender features of robots on the market and pick out four features that affect robots’ gender more, which includes decorative colors, base colors, accessories, and hair lengths. Each feature has three categories. Decorative colors are blue, pink, and achromatic color; base colors are white, light grey, and dark grey; accessories are tie, bow tie, and none; hair lengths are no hair, long hair, and short hair. Since the 3x3x3x3(81) samples are too many, we used the Taguchi’s method to turn the 81 different design combinations into 9 gender feature samples of robots. Then, according to the BSRI-12 (6 masculine adjectives; 6 feminine adjectives:) and 1 question of Semantic Different Scale (about gender perception: point 1 is more male; point 5 is more female; point 3 is neutral), we designed a questionnaire and sent out via social medias, letting the subjects fill in the Likert Scale of sample robots with their subjective feelings. On the basis of 280 questionnaires, we conducted the Quantification Theory Type I analysis and found out: (1) The visual hints of robots do affect subjects’ perception of robots’ gender role (masculine or feminine) and sex (male or female); (2) When using appearance designs as the hints of robots’ genders, subjects of different genders have different comments on robots’ gender perception (masculine or feminine, male or female), and all kinds of visual elements have different influences on male or female subjects.

並列關鍵字

Human-Robot Interaction ； Gender Cues ； Kansei Engineering ； 12-Item Bem Sex Role Inventory

參考文獻

英文文獻

1. Bainbridge, W. A., Hart, J. W., Kim, E. S., & Scassellati, B. (2011). The benefits of interactions with physically present robots over video-displayed agents. International Journal of Social Robotics, 3(1), 41-52.

2. Bem, S. L. (1974) The Measurement of Psychological Androgyny. Journal of Consulting and Clinical Psychology, 42, 2, 155-162.

3. Burgoon, J. K., & Floyd, K. (2000). Testing for the motivation impairment effect during deceptive and truthful interaction. Western Journal of Communication, 64, p243-267.

4. Carver, L. F., Vafaei, A., Guerra, R., Freire, A., & Phillips, S. P. (2013). Gender differences: Examination of the 12-item Bem Sex Role Inventory (BSRI-12) in an older Brazilian population. PloS one, 8(10).