A design method for building in-air gestural interactions

• Autor(a) principal: FIGUEIREDO, Lucas Silva
• Data de Publicação: 2017
• Tipo de documento: Tese
• Idioma: eng
• Título da fonte: Repositório Institucional da UFPE
• Texto Completo: https://repositorio.ufpe.br/handle/123456789/25630
• topic: Ciência da computação; Computação gráfica

In-air gestures are part of our everyday communications. Giving a “thumbs up,” pointing to an object of interest or raising a hand to call for attention are just a few examples. Current vision-based technologies such as the Microsoft Kinect, the Leap Motion have shown real-time tracking capabilities that enable a large range of developers and companies to explore these gestures on human-computer interactive solutions. We approach the process of building these interactions. Our efforts are to understand and aid designers, developers, and researchers in the field of Human-Computer Interaction (HCI) to build in-air gestural interfaces. With that goal in mind, we provide a set of techniques and tools to explore and prototype concepts of possible gestural interactions for a given target task. We divide the toolset into two main phases: the conception of the gestures to be used; and the prototype of solutions using these gestures. For the conception phase, we propose the use of a set of creative techniques. Moreover, we introduce a pair of web catalogs (as tools) to be used for analysis and inspiration while suggesting and creating new gestural interactions. By reviewing the literature regarding how researchers define the used gestures, we cataloged several examples according to a developed taxonomy. We also performed a similar study and built a catalog of in-air gestures present on Science Fiction (Sci-Fi) content. Sci-Fi contents, although not representing real interfaces, show potential while exploring innovative concepts that can influence the creation of new interfaces. For the prototype phase, we focused on the steps of producing and testing low-fidelity and high-fidelity prototypes for the recognition of the conceptualized gestures. For low-fidelity prototyping, we propose and validate the use of the Wizard of Oz technique, which enables fast testing of different concepts. For high-fidelity prototyping, we introduce a recognition tool called Prepose, which aims the easiness of use for creating and editing the gesture recognizers. Prepose allows a gesture to be written in natural language making it easy for developers and non-developers to read, write and edit the target gestures. At the same time, Prepose allows a gesture to be automatically written with one sample of its execution, speeding up the time to build recognizers for complex gestures. At last, we also conducted a pilot study to demonstrate the use of the toolset. In this study, we generated a set of 32 interaction concepts that was incrementally reduced while using the proposed techniques for selection and prototyping. In the end, the application of the toolset resulted in a high-fidelity prototype for the best-evaluated interaction concept.