Abstract
Dexterous grasping is the foundational primitive in embodied AI, demanding massive data to train robust models. As real-world data collection is expensive, simulation has become the mainstream paradigm. Yet, while cluttered scenes best reflect real-world applications, learning to grasp within them is bottlenecked by a critical scarcity of large-scale data. To resolve this, we curate high-quality 3D objects and supporting bases, proposing a scalable seed-and-filter strategy that bypasses sluggish scene-level optimization. This yields an unprecedented benchmark comprising over 2.6 million scenes and 0.4B scene-specific grasp ground truths, featuring diverse realistic layouts paired with rich semantic and geometric observations. Furthermore, we introduce the OmniDex model to overcome the grasp multimodality and last-millimeter precision errors plaguing current generative models. By coupling Soft Winner-Takes-All learning with human-inspired physical constraints during training, and utilizing physics-driven ranking, our approach achieves robust dexterous grasping without the latency of post-optimization. Experimental results show that OmniDex model achieves state-of-the-art performance and strong generalization across diverse scenes, views, and unseen objects.
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Fang, N., Luo, Z., Mo, Y., Huang, S., Liu, J., Liu, Y., Zhou, Z., Jin, C., Wang, X., & Li, H. (2026). OmniDex: Scaling Dexterous Hand Grasping to Diverse Cluttered Scenes. https://omanscience.com/en/articles/omnidex-scaling-dexterous-hand-grasping-to-diverse-cluttered-scenes
MLA 9
Fang, Naiyu, et al. "OmniDex: Scaling Dexterous Hand Grasping to Diverse Cluttered Scenes." https://omanscience.com/en/articles/omnidex-scaling-dexterous-hand-grasping-to-diverse-cluttered-scenes.
Chicago (author–date)
Fang, Naiyu, Zhongjin Luo, Yuxin Mo, Siyuan Huang, Jianbo Liu, Yufei Liu, Zheyuan Zhou, Chenkai Jin, Xiaogang Wang, and Hongsheng Li. 2026. "OmniDex: Scaling Dexterous Hand Grasping to Diverse Cluttered Scenes." https://omanscience.com/en/articles/omnidex-scaling-dexterous-hand-grasping-to-diverse-cluttered-scenes.
Harvard
Fang, N., Luo, Z., Mo, Y., Huang, S., Liu, J., Liu, Y., Zhou, Z., Jin, C., Wang, X. and Li, H. (2026) 'OmniDex: Scaling Dexterous Hand Grasping to Diverse Cluttered Scenes', Available at: https://omanscience.com/en/articles/omnidex-scaling-dexterous-hand-grasping-to-diverse-cluttered-scenes.
Vancouver
Fang N, Luo Z, Mo Y, Huang S, Liu J, Liu Y, et al. OmniDex: Scaling Dexterous Hand Grasping to Diverse Cluttered Scenes. https://omanscience.com/en/articles/omnidex-scaling-dexterous-hand-grasping-to-diverse-cluttered-scenes
IEEE
N. Fang, Z. Luo, Y. Mo, S. Huang, J. Liu, Y. Liu, Z. Zhou, C. Jin, X. Wang, and H. Li, "OmniDex: Scaling Dexterous Hand Grasping to Diverse Cluttered Scenes," https://omanscience.com/en/articles/omnidex-scaling-dexterous-hand-grasping-to-diverse-cluttered-scenes.