MMGNet is a Multi-Modal Grasp Synthesis Network that addresses bottlenecks through hierarchical image-to-3D feature injection and rotational graspness modeling and ablation studies show that these two modules provide complementary gains.
Abstract
Grasping is a foundational capability in robotic manipulation. Scene-level 6-DoF grasp synthesis relies on point clouds for metric geometry, yet real-world point observations are often noisy, incomplete, and over-smoothed. RGB images provide object-region, boundary, and appearance cues, yet two bottlenecks limit their effective use in scene representation and local grasp reasoning. We present MMGNet, a Multi-Modal Grasp Synthesis Network that addresses these bottlenecks through hierarchical image-to-3D feature injection and rotational graspness modeling. Hierarchical injection integrates multi-scale image features throughout 3D representation formation, while the rotational graspness jointly predicts the grasp approach direction and in-plane rotation before organizing local image-informed features through grasp-shaded grouping. Ablation studies show that these two modules provide complementary gains. MMGNet achieves a mean AP of 60.34 on GraspNet-1Billion with RealSense, surpassing all compared methods. In real-world experiments with adversarial grasp objects, MMGNet improves the grasp success rate over AnyGrasp by 8.46 percentage points. \emph{Code, models, and data will be released upon publication.}
EquiGQNet is proposed, an efficient 6-DoF grasp quality evaluator that combines the strengths of both approaches to achieve grasping performance comparable to the early-fusion baseline and substantially outperforms late fusion on objects with complex geometry and limited graspable regions.
Sungwon Seo, Jaeseog Won, Ji-You Shin et al.· 0 citations
An integrated modular approach that decouples collision reasoning, feature extraction and refinement is introduced, which significantly enhances grasping robustness and physical consistency in cluttered scenes.
Hui-Xuan Yang, Wen Wang, Wei Song et al.· Industrial robot· 0 citations
This work proposes center direction regression network (CeDiRNet)-6DoF, a deep learning framework that jointly predicts effective grasp points and the complete 6-DoF grasp pose from the observed cloth configuration by integrating dense 3-D grasp regression with segmentation and sine-cosine-encoded Euler angles.
Domen Tabernik, Peter Nimac, Jan Jerićević et al.· IEEE Transactions on Cyberne...· 0 citations
Dexterous grasping in multi-object scenes remains a fundamental challenge for robotic manipulation, particularly under single-view perception where occlusion and incomplete geometry introduce significant uncertainty. To address this challenge, we propose DGS-Net, an end-to-end dexterous grasp prediction network that di...
Tao Geng, Da-Hai Yang, Zi-Wei Liu et al.· IEEE Transactions on Automat...· 0 citations
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...
Nai-Yu Fang, Zhong-Jin Luo, Yu-Xin Mo et al.· 0 citations
Synthesizing realistic articulated hand-object interactions is a fundamental problem in virtual reality, embodied intelligence, and digital human applications. Existing methods for dexterous grasp synthesis typically regress or denoise poses in a joint space that couples global rigid motion with local articulation, whi...
Fu-Qiang Zhao, Qian Liu· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.