Human torso pose forecasting in the real world - Robotics Institute Carnegie Mellon University

Human torso pose forecasting in the real world

Workshop Paper, RSS '18 Multimodal Perception and Control Workshop, June, 2018

Abstract

In this paper, we describe a multi-modal approach to human torso pose estimation and forecasting. Our end-to-end system combines RGB images and point cloud information to reason about 3D human pose. We use a simple filter fit method to forecast torso pose. Further, we evaluate the forecasting performance quantitatively on the Human3.6M motion capture dataset and qualitatively on a furniture assembly task. Our simple forecasting algorithm outperforms complicated recurrent neural network methods, while being faster on the torso pose forecasting task.

BibTeX

@workshop{Biswas-2018-113248,
author = {Abhijat Biswas and Henny Admoni and Aaron Steinfeld},
title = {Human torso pose forecasting in the real world},
booktitle = {Proceedings of RSS '18 Multimodal Perception and Control Workshop},
year = {2018},
month = {June},
}