Research

ArtManip: Category-Level In-Hand Manipulation of Articulated Objects

BIGAI Dexterity Team

Yang Yang *  Tengyu Liu *  Puhao Li   Zeyuan Chen   Yuyang Li   Xingwan Wang   Yingying Wu   Zhaopeng Cui ✉  Siyuan Huang ✉ 

Learning dexterous policies of in-hand articulated objects that generalize across intra-class instances and initial grasps, with zero-shot sim2real capability.

RoboVerse: Towards a Unified Platform, Dataset and Benchmark for Scalable and Generalizable Robot Learning

Haoran Geng *  Feishi Wang *  Songlin Wei *  Yuyang Li *  Bangjun Wang *  Boshi An *  Charlie Tianyue Cheng *  Haozhe Lou   Peihao Li   Yen-Jen Wang   Yutong Liang   Dylan Goetting   Chaoyi Xu   Haozhe Chen   Yuxi Qian   Yiran Geng   Jiageng Mao   Weikang Wan   Mingtong Zhang   Jiangran Lyu   Siheng Zhao   Jiazhao Zhang   Jialiang Zhang   Chengyang Zhao   Haoran Lu   Yufei Ding   Ran Gong   Yuran Wang   Yuxuan Kuang   Ruihai Wu   Baoxiong Jia   Carlo Sferrazza   Hao Dong   Siyuan Huang   Koushil Sreenath   Yue Wang   Jitendra Malik   Pieter Abbeel  

A comprehensive framework comprising a simulation platform, a synthetic dataset, and unified benchmarks.

51 citations

Ag2Manip: Learning Novel Manipulation Skills with Agent-Agnostic Visual and Action Representations

Puhao Li *  Tengyu Liu *  Yuyang Li   Muzhi Han   Haoran Geng   Shu Wang   Yixin Zhu   Song-Chun Zhu   Siyuan Huang ✉ 

Learning novel manipulation skills with agent-agnostic value and action representations.

41 citations

Grasp Multiple Objects with One Hand

Yuyang Li   Bo Liu   Yiran Geng   Puhao Li   Yaodong Yang   Yixin Zhu   Tengyu Liu ✉  Siyuan Huang ✉ 

We propose MultiGrasp, a two-stage framework for simultaneous multi-object grasping with multi-finger dexterous hands. In addition, we contribute Grasp'Em, a large-scale synthetic multi-object grasping dataset.

64 citations

Towards Human-level Dexterous Teleoperation

BIGAI Dexterity Team

Puhao Li *  Zeyuan Chen *  Yingying Wu *  Pengkun Wei   Yuyang Li   Tianyu Wang   Jiaxiao Shi   Mingrui Yu   Baoxiong Jia   Song-Chun Zhu   Tengyu Liu ✉  Siyuan Huang ✉ 

TeleDexter is a learned cerebellum for dexterous robot hands — a low-level co-tracking controller that maps operator intent into contact-rich execution in real time.

0 citations

TacMan-Turbo: Proactive Tactile Control for Robust and Efficient Articulated Object Manipulation

Zihang Zhao *  Zhenghao Qi *  Yuyang Li   Leiyao Cui   Zhi Han ✉  Lecheng Ruan ✉  Yixin Zhu ✉ 

A proactive tactile control framework for articulated object manipulation.

7 citations

ManipTrans: Efficient Dexterous Bimanual Manipulation Transfer via Residual Learning

Kailin Li   Puhao Li   Tengyu Liu   Yuyang Li   Siyuan Huang ✉ 

A novel two-stage method for efficiently transferring human bimanual skills to dexterous robotic hands in simulation.

101 citations

PreAfford: Affordance-based Pre-grasping across Objects and Scenes

Kairui Ding *  Boyuan Chen   Ruihai Wu   Yuyang Li   Zongzheng Zhang   Huan-ang Gao   Siqi Li   Yixin Zhu   Guyue Zhou   Hao Dong   Hao Zhao ✉ 

A universal pre-grasping manipulation framework using affordance representation.

19 citations

GenDexGrasp: Generalizable Dexterous Grasping

Puhao Li *  Tengyu Liu *  Yuyang Li   Yiran Geng   Yaodong Yang   Yixin Zhu   Siyuan Huang ✉ 

We propose GenDexGrasp, a versatile dexterous grasping method that can generalize to out-of-domain robotic hands. In addition, we contribute MultiDex, a large-scale synthetic dexterous grasping dataset.

169 citations