Object Segmentation through Human-Robot Interactions in the Frequency Domain

Abstract

This paper presents a new embodied approach for object segmentation by a humanoid robot. It relies on interactions with a human teacher that drives the robot through the process of segmenting objects from arbitrarily complex, nonstatic images. Objects from a large spectrum of different scenarios are successfully segmented by the proposed algorithms. The paper discusses embodied object segmentation; detection of events in the frequency domain, including event detection, tracking, and multi-scale periodic detection; segmentation by passive demonstration; segmentation through active actuation; segmentation by poking; experimental results for object segmentation in terms of robustness; and conclusions and future work.