Deformation-Aware Data-Driven Grasp Synthesis

Grasp synthesis for 3-D deformable objects remains a little-explored topic, most works aiming to minimize deformations. However, deformations are not necessarily harmful—humans are, example, able exploit generate new potential grasps. How achieve that on robot is though an open question. This letter proposes approach uses object stiffness information in addition depth images synthesizing high-quality We this by incorporating as additional input state-of-the-art deep grasp planning network. also curate synthetic dataset of grasps varying using the Isaac Gym simulator training experimentally validate and compare our proposed against case where we do incorporate total 2800 simulation 560 real Franka Emika Panda. The experimental results show significant improvement success rate wide range with shapes, sizes, stiffnesses. Furthermore, demonstrate can different grasping strategies values. Together, clearly value when stiffness. Code video available at: https://irobotics.aalto.fi/defggcnn/ .