Wavelet-based Indoor Object Recognition through Human Interaction

Abstract

In this paper a preliminary work towards grounded robot concept learning through its vision system and human interaction is presented. With a lifelong learning server, introduced in previous work [7], the robot can learn to recognize instances of such concepts as "Person", "Trash-can" and "Triangle sign" from the Haar wavelet transform of its vision space based on the instruction of a human teacher. Some experimental results and comparison with our previous work are also presented.