Publicado em: 09/08/2012

Dissertação de Mestrado em Interação Humano-Computador

UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL
INSTITUTO DE INFORMÁTICA
PROGRAMA DE POS-GRADUAÇÃO EM COMPUTAÇÃO
———————————————————
DEFESA DE DISSERTAÇÃO DE MESTRADO

Aluno: Henrique Galvan Debarba
Orientadora: Profa. Dra. Luciana Porcher Nedel
Coorientador: Prof. Dr. Anderson Maciel

Título: Selection in 2D and 3D Environments with Levels of Precision and Progressive Refinement
Linha de Pesquisa: Interação Humano-Computador

Data: 13/08/2012
Hora: 16h
Local: Auditório José Mauro Volkmer de Castilho, Prédio 43424 – Instituto de Informática

Banca Examinadora:
Profa. Dra. Carla Maria Dal Sasso Freitas (UFRGS)
Prof. Dr. Felipe Gomes de Carvalho (PUC-Rio)
Prof. Dr. Marcelo Soares Pimenta (UFRGS)

Presidente da Banca: Profa. Dra. Luciana Porcher Nedel

Abstract:
Selection is one of the four fundamental forms of interaction in a virtual world. It is the ability of the user to specify objects in the virtual environment for subsequent actions. The literature is rich in immediate selection techniques; however, this class of technique is exposed to problems of selection accuracy, ambiguity and complexity. These issues can be addressed using techniques of selection by: progressive refinement, which consists of reducing the amount of selectable objects through refinement steps until the desired object is selected; or levels of precision, which consists of increasing accuracy of pointing by manipulating the control-to-display parameters as well as by the combination of pointing approaches. Levels of precision is a class of selection techniques we are proposing in this dissertation. We also present a survey that comprehends literature on these two classes of selection techniques. Furthermore, we propose the LOP-cursor and the disambiguation canvas selection techniques. The first rely on multiple levels of precision to achieve very high control over the cursor position, while the second uses the high resolution of input provided by a mobile device touchscreen to quickly disambiguate a selection among hundreds of objects. Finally, we present the two-legged cursor metaphor, a novel approach for simultaneously pointing of two distinct locations. The two-legged cursor is intended to be used to quickly perform composite tasks, such as drag and drop. User evaluation shows that our approaches are promising, and that the design space of techniques using mobile devices can lead to numerous possibilities. Users were able to select targets as little as ≈ 0.1⁰ of visual angular size with LOP-cursor. In one of its evaluations, disambiguation canvas achieved an error rate < 0.01 per selection trial. Finally, both techniques performed faster than ray-casting for small targets, and overall they were preferred by users. Keywords: 2D interaction, 3D interaction, selection, user studies.