diff --git a/.DS_Store b/.DS_Store new file mode 100644 index 0000000..42d49d6 Binary files /dev/null and b/.DS_Store differ diff --git a/index.html b/index.html index 354e47c..3f2f35b 100644 --- a/index.html +++ b/index.html @@ -1,7 +1,7 @@ - DexSing + DexSinGrasp @@ -47,12 +47,12 @@
-

Learning Synergistic Dexterous Singulation in Cluttered Environments

+

DexSinGrasp: Learning a Unified Policy for Dexterous Object Singulation and Grasping in Cluttered Environments

- Lixin Xu, Zhewei Gui, Zixuan Liu, Zeyu Jiang, Zixuan Liu,
- Jingxiang Guo, Zhixuan Xu, Chongkai Gao, Zhixuan Xu, Lin Shao + Lixin Xu, Zixuan Liu, Zhewei Gui, Jingxiang Guo,
+ Zeyu Jiang, Zhixuan Xu, Chongkai Gao, Lin Shao
@@ -64,7 +64,7 @@

Learning Synergistic Dexterous Singulat - arXiv + arXiv (Coming Soon) @@ -72,17 +72,18 @@

Learning Synergistic Dexterous Singulat - Paper + Paper (Coming Soon) - + + @@ -90,7 +91,7 @@

Learning Synergistic Dexterous Singulat - Code + Code (Coming Soon) @@ -100,7 +101,7 @@

Learning Synergistic Dexterous Singulat - Data + Dataset (Coming Soon)

@@ -116,7 +117,7 @@

Learning Synergistic Dexterous Singulat
@@ -131,34 +132,154 @@

Learning Synergistic Dexterous Singulat

Abstract

- Grasping objects in cluttered environments remains a fundamental yet challenging problem in robotic manipulation. While prior works have explored learning-based synergies for two-fingered grippers, few works leverage high degrees of freedom (DoF) of dexterous hands for synegistic singulation in cluttered environments. In this work, we propose DexSing, a synergistic dexterous singulation policy using reinforcement learning, which enables to perform singulation for grasping with high dexterity, thus significantly improving grasping efficiency in cluttered environments. We incorporate curriculum learning to enhance success rate and generalization across diverse clutter conditions and employ policy distillation to obtain a deployable vision-based grasping strategy. To evaluate our approach, we introduce a set of cluttered grasping tasks with varying object arrangements and occlusion levels. Experimental results demonstrate that our method outperforms baselines in terms of efficiency and grasping success rate, particularly in dense clutter. + Grasping objects in cluttered environments remains a fundamental yet challenging problem in robotic manipulation. While prior works have explored learning-based synergies between pushing and grasping for two-fingered grippers, few have leveraged the high degrees of freedom (DoF) in dexterous hands to perform efficient singulation for grasping in cluttered settings. + In this work, we introduce DexSinGrasp, a unified policy for dexterous object singulation and grasping. DexSinGrasp enables high-dexterity object singulation to facilitate grasping, significantly improving efficiency and effectiveness in cluttered environments. We incorporate curriculum learning to enhance success rates and generalization across diverse clutter conditions, while policy distillation enables a deployable vision-based grasping strategy. + To evaluate our approach, we introduce a set of cluttered grasping tasks with varying object arrangements and occlusion levels. Experimental results show that our method outperforms baselines in both efficiency and grasping success rate. Experimental results show that our method outperforms baselines in both efficiency and grasping success rate, particularly in dense clutter. Codes, appendix, and videos are available on our project website https://nus-lins-lab.github.io/dexsingweb/.

+ + + - + + + +
+
+

Cluttered Environments Generation

+
+
+ Overview of DexSinGrasp framework image +
Dense and random arrangement settings. We introduce a cluttered + environment generation module to create diverse object settings, including + different obstacle quantities from 0 to 8, dense and random arrangements, + various poses, and block shapes. For simplicity, we only show obstacles + with numbers 4 to 8.
+
+
+
+

+ +

+
+
+
+ + + + +
+
+

Simulation Experiments

+
+ +
+
+
+ Dense 4 objects +
Dense arrangement with 4 obstacles
+
+
+
+
+ Dense 6 objects +
Dense arrangement with 6 obstacles
+
+
+
+
+ Dense 8 objects +
Dense arrangement with 8 obstacles
+
+
+
+ +
+
+
+ Random 4 objects +
Random arrangement with 4 obstacles
+
+
+
+
+ Random 6 objects +
Random arrangement with 6 obstacles
+
+
+
+
+ Random 8 objects +
Random arrangement with 8 obstacles
+
+
+
+
+
+

+ +

+
+
+
+ + + +
+
+

Real-World Experiments

+
+
+ Overview of DexSinGrasp framework image +
We conduct real-world experiments using an xArm6 equipped with a LEAP Hand. Results show the efficacy of our vision-based student policy.
+
+
+
+

+ +

+
+
+
+ + + + + + +
+
+

BibTeX

+ 
@article{xu2025dexsingrasp,
+    title={DexSinGrasp: Learning a Unified Policy for Dexterous Object Singulation and Grasping in Cluttered Environments},
+    author={Xu, Lixin and Liu, Zixuan and Gui, Zhewei and Guo, Jingxiang and Jiang, Zeyu and Xu, Zhixuan and Gao, Chongkai and Shao, Lin},
+    journal={arXiv preprint},
+    year={2025}
+}
- - - - -