Harnessing multi-user design and computation to devise archetypal whole-of-body gestures: a novel framework

  • Authors:
  • Suranjith De Silva;Michael Barlow;Adam Easton

  • Affiliations:
  • University of New South Wales, Australia;University of New South Wales, Australia;Simcentric Technologies

  • Venue:
  • Proceedings of the 25th Australian Computer-Human Interaction Conference: Augmentation, Application, Innovation, Collaboration
  • Year:
  • 2013

Quantified Score

Hi-index 0.00

Visualization

Abstract

A novel framework is proposed to capture the variability in end user designed gestures and extract archetypal patterns from a pool of gestures sourced from multiple participants. The primary objective is to identify a gesture library that is preferred by the end user population so as to control a human avatar in a 3D virtual environment using whole-of-body gestures. By adapting a group based user centric study, different gesture designs from 36 participants were elicited. Analysis shows that the existing techniques are incapable of extracting archetypal patterns in gestures from such an unconstrained gesture space. As such, a formal notation, followed by hierarchical clustering, is used to provide an abstract representation of gesture designs and then to distil the archetypal gesture patterns from the pool of highly variable gestures. User acceptance of the extracted gestures was performed for validation and common motion patterns were identified from the provided user ratings. The gesture library selected by the framework is compared against the gesture library extracted based on the user ranking, and the similarities and differences between two gesture libraries are presented.