ImitationNet: Unsupervised Human-to-Robot Motion Retargeting via Shared Latent Space

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Yan, Yashuai, Mascaro, Esteve Valls, Lee, Dongheui
Format: Preprint
Veröffentlicht: 2023
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866916197017059328
author Yan, Yashuai
Mascaro, Esteve Valls
Lee, Dongheui
author_facet Yan, Yashuai
Mascaro, Esteve Valls
Lee, Dongheui
contents This paper introduces a novel deep-learning approach for human-to-robot motion retargeting, enabling robots to mimic human poses accurately. Contrary to prior deep-learning-based works, our method does not require paired human-to-robot data, which facilitates its translation to new robots. First, we construct a shared latent space between humans and robots via adaptive contrastive learning that takes advantage of a proposed cross-domain similarity metric between the human and robot poses. Additionally, we propose a consistency term to build a common latent space that captures the similarity of the poses with precision while allowing direct robot motion control from the latent space. For instance, we can generate in-between motion through simple linear interpolation between two projected human poses. We conduct a comprehensive evaluation of robot control from diverse modalities (i.e., texts, RGB videos, and key poses), which facilitates robot control for non-expert users. Our model outperforms existing works regarding human-to-robot retargeting in terms of efficiency and precision. Finally, we implemented our method in a real robot with self-collision avoidance through a whole-body controller to showcase the effectiveness of our approach. More information on our website https://evm7.github.io/UnsH2R/
format Preprint
id arxiv_https___arxiv_org_abs_2309_05310
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle ImitationNet: Unsupervised Human-to-Robot Motion Retargeting via Shared Latent Space
Yan, Yashuai
Mascaro, Esteve Valls
Lee, Dongheui
Robotics
Artificial Intelligence
This paper introduces a novel deep-learning approach for human-to-robot motion retargeting, enabling robots to mimic human poses accurately. Contrary to prior deep-learning-based works, our method does not require paired human-to-robot data, which facilitates its translation to new robots. First, we construct a shared latent space between humans and robots via adaptive contrastive learning that takes advantage of a proposed cross-domain similarity metric between the human and robot poses. Additionally, we propose a consistency term to build a common latent space that captures the similarity of the poses with precision while allowing direct robot motion control from the latent space. For instance, we can generate in-between motion through simple linear interpolation between two projected human poses. We conduct a comprehensive evaluation of robot control from diverse modalities (i.e., texts, RGB videos, and key poses), which facilitates robot control for non-expert users. Our model outperforms existing works regarding human-to-robot retargeting in terms of efficiency and precision. Finally, we implemented our method in a real robot with self-collision avoidance through a whole-body controller to showcase the effectiveness of our approach. More information on our website https://evm7.github.io/UnsH2R/
title ImitationNet: Unsupervised Human-to-Robot Motion Retargeting via Shared Latent Space
topic Robotics
Artificial Intelligence
url https://arxiv.org/abs/2309.05310