3D Human Pose Estimation in RGBD Images for Robotic Task Learning

TL;DR

This paper introduces a method to accurately estimate 3D human poses from RGBD images, enabling robots to learn manipulation tasks by observing humans without markers, outperforming monocular approaches.

Contribution

It presents a novel approach combining RGB and depth data for real-world 3D human pose estimation, facilitating markerless robot learning from demonstration.

Findings

Outperforms monocular 3D pose estimation methods.

Enables robots to imitate human manipulation actions.

Works effectively in real-world scenarios.

Abstract

We propose an approach to estimate 3D human pose in real world units from a single RGBD image and show that it exceeds performance of monocular 3D pose estimation approaches from color as well as pose estimation exclusively from depth. Our approach builds on robust human keypoint detectors for color images and incorporates depth for lifting into 3D. We combine the system with our learning from demonstration framework to instruct a service robot without the need of markers. Experiments in real world settings demonstrate that our approach enables a PR2 robot to imitate manipulation actions observed from a human teacher.