GenDexHand: Generative Simulation for Dexterous Hands

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Feng, Xu, Zhuxiu, Chu, Tianzhe, Zhou, Xunzhe, Sun, Li, Wu, Zewen, Gao, Shenghua, Li, Zhongyu, Yang, Yanchao, Ma, Yi
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909884724805632
author Chen, Feng
Xu, Zhuxiu
Chu, Tianzhe
Zhou, Xunzhe
Sun, Li
Wu, Zewen
Gao, Shenghua
Li, Zhongyu
Yang, Yanchao
Ma, Yi
author_facet Chen, Feng
Xu, Zhuxiu
Chu, Tianzhe
Zhou, Xunzhe
Sun, Li
Wu, Zewen
Gao, Shenghua
Li, Zhongyu
Yang, Yanchao
Ma, Yi
contents Data scarcity remains a fundamental bottleneck for embodied intelligence. Existing approaches use large language models (LLMs) to automate gripper-based simulation generation, but they transfer poorly to dexterous manipulation, which demands more specialized environment design. Meanwhile, dexterous manipulation tasks are inherently more difficult due to their higher degrees of freedom. Massively generating feasible and trainable dexterous hand tasks remains an open challenge. To this end, we present GenDexHand, a generative simulation pipeline that autonomously produces diverse robotic tasks and environments for dexterous manipulation. GenDexHand introduces a closed-loop refinement process that adjusts object placements and scales based on vision-language model (VLM) feedback, substantially improving the average quality of generated environments. Each task is further decomposed into sub-tasks to enable sequential reinforcement learning, reducing training time and increasing success rates. Our work provides a viable path toward scalable training of diverse dexterous hand behaviors in embodied intelligence by offering a simulation-based solution to synthetic data generation. Our website: https://winniechen2002.github.io/GenDexHand/.
format Preprint
id arxiv_https___arxiv_org_abs_2511_01791
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle GenDexHand: Generative Simulation for Dexterous Hands
Chen, Feng
Xu, Zhuxiu
Chu, Tianzhe
Zhou, Xunzhe
Sun, Li
Wu, Zewen
Gao, Shenghua
Li, Zhongyu
Yang, Yanchao
Ma, Yi
Robotics
Artificial Intelligence
Data scarcity remains a fundamental bottleneck for embodied intelligence. Existing approaches use large language models (LLMs) to automate gripper-based simulation generation, but they transfer poorly to dexterous manipulation, which demands more specialized environment design. Meanwhile, dexterous manipulation tasks are inherently more difficult due to their higher degrees of freedom. Massively generating feasible and trainable dexterous hand tasks remains an open challenge. To this end, we present GenDexHand, a generative simulation pipeline that autonomously produces diverse robotic tasks and environments for dexterous manipulation. GenDexHand introduces a closed-loop refinement process that adjusts object placements and scales based on vision-language model (VLM) feedback, substantially improving the average quality of generated environments. Each task is further decomposed into sub-tasks to enable sequential reinforcement learning, reducing training time and increasing success rates. Our work provides a viable path toward scalable training of diverse dexterous hand behaviors in embodied intelligence by offering a simulation-based solution to synthetic data generation. Our website: https://winniechen2002.github.io/GenDexHand/.
title GenDexHand: Generative Simulation for Dexterous Hands
topic Robotics
Artificial Intelligence
url https://arxiv.org/abs/2511.01791