Harnessing Human Manipulation - Robotics Institute Carnegie Mellon University

Harnessing Human Manipulation

Workshop Paper, NSF / ARL Workshop on Cloud Robotics: Challenges and Opportunities, February, 2013

Abstract

This paper describes a new project to analyze the vast corpus of human manipulation videos available on the web, using citizen scientists. Humans understand human manipulation well enough to segment videos appropriately, to assign intent, and to recognize actions even from poor imagery. A large corpus of analyzed human manipulation would add to our understanding of manipulation, support automated analysis, facilitate collaborative manipulation and generally accelerate robotic manipulation research. The paper presents our initial results with a web interface used by nine “citizen scientists” to analyze sandwich making.

BibTeX

@workshop{Mason-2013-7669,
author = {Matthew T. Mason and Nancy Pollard and Alberto Rodriguez and Ryan Kerwin},
title = {Harnessing Human Manipulation},
booktitle = {Proceedings of NSF / ARL Workshop on Cloud Robotics: Challenges and Opportunities},
year = {2013},
month = {February},
}