Cross-Embodiment Dexterous Grasping with Reinforcement Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yuan, Haoqi, Zhou, Bohan, Fu, Yuhui, Lu, Zongqing
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929525619687424
author Yuan, Haoqi
Zhou, Bohan
Fu, Yuhui
Lu, Zongqing
author_facet Yuan, Haoqi
Zhou, Bohan
Fu, Yuhui
Lu, Zongqing
contents Dexterous hands exhibit significant potential for complex real-world grasping tasks. While recent studies have primarily focused on learning policies for specific robotic hands, the development of a universal policy that controls diverse dexterous hands remains largely unexplored. In this work, we study the learning of cross-embodiment dexterous grasping policies using reinforcement learning (RL). Inspired by the capability of human hands to control various dexterous hands through teleoperation, we propose a universal action space based on the human hand's eigengrasps. The policy outputs eigengrasp actions that are then converted into specific joint actions for each robot hand through a retargeting mapping. We simplify the robot hand's proprioception to include only the positions of fingertips and the palm, offering a unified observation space across different robot hands. Our approach demonstrates an 80% success rate in grasping objects from the YCB dataset across four distinct embodiments using a single vision-based policy. Additionally, our policy exhibits zero-shot generalization to two previously unseen embodiments and significant improvement in efficient finetuning. For further details and videos, visit our project page https://sites.google.com/view/crossdex.
format Preprint
id arxiv_https___arxiv_org_abs_2410_02479
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Cross-Embodiment Dexterous Grasping with Reinforcement Learning
Yuan, Haoqi
Zhou, Bohan
Fu, Yuhui
Lu, Zongqing
Robotics
Machine Learning
Dexterous hands exhibit significant potential for complex real-world grasping tasks. While recent studies have primarily focused on learning policies for specific robotic hands, the development of a universal policy that controls diverse dexterous hands remains largely unexplored. In this work, we study the learning of cross-embodiment dexterous grasping policies using reinforcement learning (RL). Inspired by the capability of human hands to control various dexterous hands through teleoperation, we propose a universal action space based on the human hand's eigengrasps. The policy outputs eigengrasp actions that are then converted into specific joint actions for each robot hand through a retargeting mapping. We simplify the robot hand's proprioception to include only the positions of fingertips and the palm, offering a unified observation space across different robot hands. Our approach demonstrates an 80% success rate in grasping objects from the YCB dataset across four distinct embodiments using a single vision-based policy. Additionally, our policy exhibits zero-shot generalization to two previously unseen embodiments and significant improvement in efficient finetuning. For further details and videos, visit our project page https://sites.google.com/view/crossdex.
title Cross-Embodiment Dexterous Grasping with Reinforcement Learning
topic Robotics
Machine Learning
url https://arxiv.org/abs/2410.02479