A study for the identification of a full-body gesture language for enabling natural user interaction

Abstract

Most proposals addressing gesture recognition consider two main stages: gesture identification in a source and association between gestures and meanings. About the semantic meaning of gestures, it is important to consider that due to their cultural and linguistic specificity each of them may be mapped to many concepts and vice versa, making them ambiguous and incompletely defined. From the HCI perspective, there is work in the literature presenting elicitation studies on which researchers find gestures sets to allow user interaction with tailored applications under determined contexts. In this paper, we present a full-body language for enabling gesture-based interaction in different contexts. For this purpose, 70 users were asked to provide the gestures they would use as commands within different existing applications, keeping awareness of the relationship between the tasks they were asked to do on the applications and abstract tasks, resulting on 980 gestures which were compared for generating a reduced set of 68 gestures consisting on their graphic representation, a textual description, an anthropometric characterizations for each of them, and generic labels. Interpretation and insights obtained during the experiment are also reported.