analiza aiqbench
EquiDexFlow advances dexterous manipulation by integrating physical constraints like Coulomb friction directly into an SE(3)-equivariant generative flow, eliminating the need for downstream verification steps common in current grasp models. The approach successfully demonstrates zero joint deviation and stable pick-and-hold performance across asymmetric objects on both Allegro and LEAP robotic hands using only open-loop control. This work represents a significant step toward physically robust AI agents capable of handling complex real-world grasping tasks without relying on post-processing corrections.
EquiDexFlow: Contact-Grounded SE(3)-Equivariant Dexterous Grasp Generative Flows
arXiv:2606.12728v2 Announce Type: replace Abstract: Most learned dexterous grasp generators relegate contact forces to a downstream verification step, so a kinematically-plausible pose can still violate the conditions for a stable physical grasp. We address this with EquiDexFlow, an SE(3)-equivariant flow-matching model that jointly predicts wrist pose, joint angles, fingertip contacts, surface normals, and contact forces from an object point cloud. Our architecture projects contacts onto the object surface and forces into the Coulomb friction cone by construction, so placement and friction compliance hold without loss penalties. We prove end-to-end SE(3) equivariance and verify it empirically over 200 rotations, with wrist residuals below $0.04^\circ$ and exactly zero joint deviation. Trained on 8,100 force-closure grasps across 81 objects for the 16-DoF Allegro Hand, our model achieves zero friction violations, the best composite score, and the lowest wrench residual among all ablation variants. We retarget decoded fingertip contacts to a 16-DoF LEAP Hand via per-finger inverse kinematics, and our hardware-feasible refinement places every joint at least 5% inside its actuator envelope while preserving wrench balance. On the physical robot, retargeted EquiDexFlow-decoded grasps complete open-loop pick-and-hold trials on all six test objects, with every asymmetric object succeeding at both the canonical pose and a $120^\circ$ co-rotation. Videos, code, and checkpoints are available at https://equidexflow.github.io.
Abstract:Most learned dexterous grasp generators relegate contact forces to a downstream verification step, so a kinematically-plausible pose can still violate the conditions for a stable physical grasp. We address this with EquiDexFlow, an SE(3)-equivariant flow-matching model that jointly predicts wrist pose, joint angles, fingertip contacts, surface normals, and contact forces from an object point cloud. Our architecture projects contacts onto the object surface and forces into the Coulomb friction cone by construction, so placement and friction compliance hold without loss penalties. We prove end-to-end SE(3) equivariance and verify it empirically over 200 rotations, with wrist residuals below $0.04^\circ$ and exactly zero joint deviation. Trained on 8,100 force-closure grasps across 81 objects for the 16-DoF Allegro Hand, our model achieves zero friction violations, the best composite score, and the lowest wrench residual among all ablation variants. We retarget decoded fingertip contacts to a 16-DoF LEAP Hand via per-finger inverse kinematics, and our hardware-feasible refinement places every joint at least 5% inside its actuator envelope while preserving wrench balance. On the physical robot, retargeted EquiDexFlow-decoded grasps complete open-loop pick-and-hold trials on all six test objects, with every asymmetric object succeeding at both the canonical pose and a $120^\circ$ co-rotation. Videos, code, and checkpoints are available at this https URL.
22 pages, 11 figures, 11 tables. Project page with videos, code, and checkpoints: this https URL
Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
arXiv:2606.12728 [cs.RO]
(or arXiv:2606.12728v2 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2606.12728 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 323 cuvinte
Subiecte conexe
EquiDexFlow: Contact-Grounded SE(3)-Equivariant Dexterous Grasp Generative Flows
arXiv · 12 iun. 2026
KPGrasp: Scalable Keypoint Flow Matching for Dexterous Grasp Generation
arXiv · 9 iun. 2026
KPGrasp: Scalable Keypoint Flow Matching for Dexterous Grasp Generation
arXiv · 14 oct. 2026
GOAG: Generative and Object-Agnostic Grasp Planner for Dexterous Robotic Manipulation
arXiv · 21 aug. 2026