Learning Generalizable Hand-Object Tracking from Synthetic Demonstrations

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Yinhuai, Yu, Runyi, Tsui, Hok Wai, Lin, Xiaoyi, Zhang, Hui, Zhao, Qihan, Fan, Ke, Li, Miao, Song, Jie, Wang, Jingbo, Chen, Qifeng, Tan, Ping
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911332789387264
author Wang, Yinhuai
Yu, Runyi
Tsui, Hok Wai
Lin, Xiaoyi
Zhang, Hui
Zhao, Qihan
Fan, Ke
Li, Miao
Song, Jie
Wang, Jingbo
Chen, Qifeng
Tan, Ping
author_facet Wang, Yinhuai
Yu, Runyi
Tsui, Hok Wai
Lin, Xiaoyi
Zhang, Hui
Zhao, Qihan
Fan, Ke
Li, Miao
Song, Jie
Wang, Jingbo
Chen, Qifeng
Tan, Ping
contents We present a system for learning generalizable hand-object tracking controllers purely from synthetic data, without requiring any human demonstrations. Our approach makes two key contributions: (1) HOP, a Hand-Object Planner, which can synthesize diverse hand-object trajectories; and (2) HOT, a Hand-Object Tracker that bridges synthetic-to-physical transfer through reinforcement learning and interaction imitation learning, delivering a generalizable controller conditioned on target hand-object states. Our method extends to diverse object shapes and hand morphologies. Through extensive evaluations, we show that our approach enables dexterous hands to track challenging, long-horizon sequences including object re-arrangement and agile in-hand reorientation. These results represent a significant step toward scalable foundation controllers for manipulation that can learn entirely from synthetic data, breaking the data bottleneck that has long constrained progress in dexterous manipulation.
format Preprint
id arxiv_https___arxiv_org_abs_2512_19583
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learning Generalizable Hand-Object Tracking from Synthetic Demonstrations
Wang, Yinhuai
Yu, Runyi
Tsui, Hok Wai
Lin, Xiaoyi
Zhang, Hui
Zhao, Qihan
Fan, Ke
Li, Miao
Song, Jie
Wang, Jingbo
Chen, Qifeng
Tan, Ping
Robotics
Graphics
We present a system for learning generalizable hand-object tracking controllers purely from synthetic data, without requiring any human demonstrations. Our approach makes two key contributions: (1) HOP, a Hand-Object Planner, which can synthesize diverse hand-object trajectories; and (2) HOT, a Hand-Object Tracker that bridges synthetic-to-physical transfer through reinforcement learning and interaction imitation learning, delivering a generalizable controller conditioned on target hand-object states. Our method extends to diverse object shapes and hand morphologies. Through extensive evaluations, we show that our approach enables dexterous hands to track challenging, long-horizon sequences including object re-arrangement and agile in-hand reorientation. These results represent a significant step toward scalable foundation controllers for manipulation that can learn entirely from synthetic data, breaking the data bottleneck that has long constrained progress in dexterous manipulation.
title Learning Generalizable Hand-Object Tracking from Synthetic Demonstrations
topic Robotics
Graphics
url https://arxiv.org/abs/2512.19583