Fuzzy ensembles for embedding adaptive behaviours in semi-autonomous avatars in 3D virtual worlds

Author

Wall, Julie ; Izquierdo, Ebroul ; Qianni Zhang

Author_Institution

Multimedia & Vision Res. Group, Univ. of London, London, UK

fYear

2013

fDate

1-3 July 2013

Firstpage

1

Lastpage

6

Abstract

Semi-autonomous avatars should be both realistic and believable. The goal is to learn from and reproduce the behaviours of the user-controlled input to enable semi-autonomous avatars to plausibly interact with their human-controlled counterparts. A powerful tool for embedding autonomous behaviour is learning by imitation. Hence, in this paper an ensemble of fuzzy inference systems cluster the user input data to identify natural groupings within the data to describe the users movement and actions in a more abstract way. Multiple clustering algorithms are investigated along with a neuro-fuzzy classifier; and an ensemble of fuzzy systems are evaluated.

Keywords

avatars; fuzzy neural nets; fuzzy reasoning; human computer interaction; learning by example; pattern classification; pattern clustering; 3D virtual worlds; adaptive behaviour embedding; avatar interaction; behaviour learning; clustering algorithms; data natural grouping identification; fuzzy inference systems; fuzzy system ensemble; human-controlled counterparts; learning by imitation; neuro-fuzzy classifier; semiautonomous avatar; user action; user input data clustering; user movement; user-controlled input; Avatars; Clustering algorithms; Fuzzy logic; Fuzzy systems; Testing; Three-dimensional displays; Training; fuzzy ensembles; fuzzy inference system; semi-autonomous avatar; virtual environments;

fLanguage

English

Publisher

ieee

Conference_Titel

Digital Signal Processing (DSP), 2013 18th International Conference on

Conference_Location

Fira

ISSN

1546-1874

Type

conf

DOI

10.1109/ICDSP.2013.6622818

Filename

6622818