Sensory and a priori information integration for object pointing in interactive robot operation

Abstract

An information integration method is presented that combines sensed environmental data with a priori information to control object pointing in interactive robot operation. To identify the target object from the environmental image, a cooperative processing mechanism is introduced. This mechanism uses a distributed cooperative system for location-posture analysis of the object pattern and structural analysis of the object shape. The heterogeneous sensory information thus obtained is merged consistently and integrated in the instance frame of the work space model. The proposed method is experimentally verified using a prototype system.