EAGG: Embodiment-Aligned Grasp Generation via Geometry-Aware Graph Conditioning

EAGG (Embodiment-Aligned Grasp Generation) introduces a unified model for cross-end-effector grasp generation that utilizes a topology-aware end-effector graph and an embodiment-specific control space to enhance generalization across various grasping embodiments. The model achieves an average success rate of 56.17% on the MultiGripperGrasp benchmark, closely matching specialized training results, while significantly reducing median contact distance from 0.239 cm to 0.189 cm through iterative geometry injection. This approach demonstrates that aligning embodiment structures within a shared generator improves transferability and performance, making it a valuable tool for practitioners working with diverse robotic grippers.