|
Bielefeld University Faculty of Linguistics and Literature Faculty of Technology |
Abstract
An important prerequisite for the natural interaction of humans with a mobile robot is the robot's capability to detect potential communication partners. In this paper we present an approach which uses a combination of person recognition and tracking with sound source localization realized in a multi-modal anchoring framework. As in open environments several potential communication partners can be present simultaneously we developed a rule-based method for selecting one specific person as the current communication partner.
|
Last Change: 2003-12-12 gk256www@techfak.uni-bielefeld.de |